OPINION OF THE COURT

AMBRO, Circuit Judge.

Glenn Flemming was sentenced in February 2005 to 175 months' imprisonment for federal firearm and crack cocaine offenses committed in 2002. After the United States Sentencing Commission retroactively lowered the offense levels for most crack cocaine offenses by two levels, Flemming moved for a reduction of sentence under 18 U.S.C. § 3582(c)(2). The District Court denied his motion, concluding that it lacked authority to reduce Flemming's sentence because he was a career offender under U.S.S.G. § 4B1.1. On appeal, Flemming argues that, despite his status as a career offender, he is eligible for a sentence reduction under § 3582(c)(2) because the District Court granted him a downward departure under U.S.S.G. § 4A1.3 after concluding that the career offender enhancement overstated the seriousness of his criminal history, and instead sentenced him within the Guidelines range for crack cocaine offenses.

The narrow issue presented in this case—whether a career offender who receives a § 4A1.3 downward departure under a pre-2003 edition of the Sentencing Guidelines to the Guidelines range for crack cocaine offenses is eligible for a sentence reduction under § 3852(c)(2)—is one of first impression in our Court, but one that has divided our sister circuit courts. The First and Second Circuits, as well as a divided panel of the Fourth Circuit, have concluded that such a defendant is eligible for a sentence reduction. The Eighth and Tenth Circuits, as well as a divided panel of the Sixth Circuit, have disagreed.

Though we do so through a somewhat different analysis, we join the First, Second, and Fourth Circuit Courts in concluding that such a defendant is eligible for a sentence reduction under § 3582(c)(2). Accordingly, we vacate the District Court's order and remand for further proceedings.

I. Background

In March 2003, a federal grand jury returned an indictment charging that Flemming possessed with intent to distribute crack cocaine, in violation of 21 U.S.C. § 841(a) and (b)(1)(C) (Count One); possessed a firearm in furtherance of a drug trafficking crime, in violation of 18 U.S.C. § 924(c)(1) (Count Two); and possessed a firearm as a felon, in violation of 18 U.S.C. § 922(g) (Count Three). A jury convicted Flemming on all three counts.

A. Sentencing

Using the 2001 edition of the Sentencing Guidelines,1 the Probation Office determined that under the drug quantity table, U.S.S.G. § 2D1.1(c) (the "Crack Cocaine Guidelines"), Flemming's base offense level was 24 because he possessed more than four but less than five grams of crack cocaine. With a criminal history category of V, Flemming's Guidelines range would have been 92 to 115 months' imprisonment. However, Flemming also qualified as a career offender under U.S.S.G. § 4B1.12 (the "Career Offender Guidelines") based on two prior convictions for controlled substance offenses.3 The career offender enhancement increased Flemming's offense level to 34 and his criminal history category to VI, with a resulting Sentencing Guidelines range of 262 to 327 months' imprisonment. Flemming also faced a mandatory consecutive sentence of 60 months' imprisonment on Count Two. See 18 U.S.C. § 924(c)(1)(A)(i).

At sentencing, Flemming argued that the career offender enhancement overstated his criminal history, warranting a downward departure pursuant to U.S.S.G. § 4A1.3. Under the 2001 edition of the Sentencing Guidelines, § 4A1.3 in relevant part provided:

U.S.S.G. § 4A1.3 (2001). In United States v. Shoupe, 35 F.3d 835 (3d Cir.1994), we held that § 4A1.3 authorized a court to depart both horizontally (in criminal history category) and vertically (in offense level). Id. at 839. In the specific context of a defendant subject to the career offender enhancement, we reasoned that "[b]ecause career offender status enhances both a defendant's criminal history category and offense level, . . . a sentencing court may depart in both under the proper circumstances." Id. at 838.

Consistent with our interpretation of § 4A1.3 in Shoupe, the District Court granted Flemming a downward departure in both offense level and criminal history category. The Court explained:

Though it did not expressly quantify the extent of its departure, the offense level and criminal history category that the Court was "le[ft] . . . with" were the same offense level and criminal history category that applied under the Crack Cocaine Guidelines without the career offender enhancement (i.e., level 24 and category V). To repeat, the resulting Guidelines range was 92 to 115 months' imprisonment.

Noting that the District Court had "determined that [it was] not going to apply the career offender provisions," the Government recommended a sentence at the top of the Guidelines range. The District Court followed suit and sentenced Flemming to 115 months' imprisonment on Counts One and Three, and a mandatory consecutive sentence of 60 months' imprisonment on Count Two. Flemming's total sentence was thus 175 months' imprisonment.4

B. Motion for Sentence Reduction

Effective November 1, 2007, the Sentencing Commission issued Amendment 706 to the Guidelines, which lowered the base offense level for offenses involving most quantities of crack cocaine by two levels. U.S.S.G. app. C, amend. 706 (Nov. 1, 2007). In May 2008, the Commission made Amendment 706 retroactive. U.S.S.G. app. C, amend. 713 (Supp. May 1, 2008); U.S.S.G. § 1B1.10(c). On the basis of the Amendment, Flemming filed a pro se motion for a reduction of sentence pursuant to 18 U.S.C. § 3582(c)(2). The District Court appointed counsel to represent Flemming, and that counsel filed a supplemental brief on Flemming's behalf.

The District Court denied Flemming's motion. It reasoned that, although it had granted Flemming a downward departure under § 4A1.3, it had nonetheless adopted the findings of the Presentence Investigation Report and determined that Flemming qualified as a career offender. Citing our decision in United States v. Mateo, 560 F.3d 152 (3d Cir.2009), the Court held that Flemming was not entitled to a sentence reduction because Amendment 706 did not affect his sentencing range under the Career Offender Guidelines.5 Flemming timely appealed.

II. Jurisdiction

The District Court had jurisdiction under 18 U.S.C. § 3231. We have appellate jurisdiction under 28 U.S.C. § 1291.

II. Discussion

"[A] judgment of conviction that includes [a term of imprisonment] constitutes a final judgment," 18 U.S.C. § 3582(b), and generally may not be modified by a district court "once it has been imposed," id. § 3582(c). However, there is a limited exception to this general rule of finality:

Id. § 3582(c)(2).

We have interpreted this provision as authorizing a district court to reduce a sentence already imposed where two requirements are satisfied: (1) the defendant's initial sentence must have been "based on a sentencing range that has subsequently been lowered by the Sentencing Commission," and (2) the sentence reduction must be "consistent with applicable policy statements issued by the Sentencing Commission." Id.; United States v. Doe, 564 F.3d 305, 309 (3d Cir.2009). If these two requirements are satisfied, a court may then exercise its discretion— "after considering the factors set forth in section 3553(a) to the extent that they are applicable," 18 U.S.C. § 3582(c)(2), as well as the factors contained in the commentary to the Commission's policy statements, U.S.S.G. § 1B1.10 cmt. n. 1(B)—to determine whether a reduction in sentence is warranted (as well as the extent of any such reduction).

In this case, the District Court determined that it lacked authority to consider whether a sentence reduction was warranted because Flemming did not satisfy either of the two statutory requirements in § 3582(c)(2). Thus, we are not called on to review the District Court's exercise of its discretion, but to determine whether it was correct that Flemming is statutorily ineligible for a sentence reduction under § 3582(c)(2). We review de novo a district court's interpretation of § 3582(c)(2), as well as its interpretation of the Sentencing Guidelines. See Mateo, 560 F.3d at 154.

A. Was Flemming sentenced "based on a sentencing range" that has been lowered by Amendment 706?

Consistent with the District Court's reasoning, the Government contends that our decision in Mateo compels the conclusion that Flemming was not sentenced "based on a sentencing range" that was lowered by Amendment 706 because his Guidelines range was calculated under the Career Offender Guidelines in § 4B1.1, rather than the Crack Cocaine Guidelines in § 2D1.1(c). We disagree.

In Mateo, we held that a career offender, who received no downward departures and was sentenced within the Career Offender Guidelines range, was not eligible for a reduction in sentence even though his base offense level under the Crack Cocaine Guidelines had been subsequently lowered by Amendment 706. 560 F.3d at 155. We rejected the defendant's argument that his sentence was "based on" the Crack Cocaine Guidelines in § 2D1.1(c) simply because "the District Court consulted that section in calculating his offense level." Id. In addition, we emphasized the language from § 3582(c)(2) that the sentence imposed must have been "`based on a sentencing range that has subsequently been lowered by the Sentencing Commission,'" id. (quoting 18 U.S.C. § 3582(c)(2)) (emphasis in original), and reasoned that the term "sentencing range" "`clearly contemplates the end result of the overall guideline calculus, not the series of tentative results reached at various interim steps in the performance of that calculus.'" Id. (quoting United States v. Caraballo, 552 F.3d 6, 10 (1st Cir.2008)) (emphasis added). Thus, "`if an amended guideline does not have the effect of lowering the sentencing range actually used at sentencing, the defendant's sentence was not based on that range within the intendment of the statute.'" Id. (quoting Caraballo, 552 F.3d at 10) (emphasis added).

Far from compelling the conclusion that he is ineligible for a sentence reduction, Mateo supports Flemming's argument that his sentence was "based on" the sentencing range calculated under the Crack Cocaine Guidelines. Though the District Court agreed that Flemming technically qualified as a career offender, it declined to sentence him within that range, and instead applied the Crack Cocaine Guidelines range after determining under § 4A1.3 that the career offender enhancement overstated the seriousness of his criminal history. In other words, the District Court "actually used" the Crack Cocaine Guidelines range, rather than the Career Offender Guidelines range, when it sentenced Flemming.

The First, Second, and Fourth Circuit Courts—as well as numerous district courts, including several in our own Circuit6 —have reached the same conclusion in nearly identical circumstances.7 In United States v. McGee, 553 F.3d 225 (2d Cir.2009) (per curiam), the District Court granted a career offender a downward departure under § 4A1.3, stating that it was departing from the Career Offender Guidelines range "to the level that the defendant would have been in absent the career offender status calculation." Id. at 227 (internal quotation marks omitted). Accordingly, the Court applied the sentencing range calculated under the Crack Cocaine Guidelines and sentenced the defendant within that range. Id. The Second Circuit concluded that the defendant was eligible for a sentence reduction, as it was "apparent that [the defendant] was sentenced `based on' a sentencing guideline range that was subsequently lowered by the Sentencing Commission because the district court premised [its] ultimate sentence on the crack cocaine guidelines." Id. Indeed, because the district court "sentenced [the defendant] based on the crack cocaine guidelines[, it] would likely have considered a different sentence from the one imposed if the applicable crack guidelines had so provided." Id. at 228. Importantly, the Second Circuit perceived no conflict between the result it reached in McGee and the conclusion, consistent with Mateo, that "a defendant who is sentenced as a career offender, but does not receive a downward departure, [does not] qualif[y] for a reduced sentence." Id. at 227; see also United States v. Martinez, 572 F.3d 82, 84 (2d Cir.2009) (distinguishing the defendant in McGee—who "could have been sentenced under § 4B1.1 but was in fact sentenced under [the Crack Cocaine Guidelines in] § 2D1.1"—from a defendant who is sentenced within the Career Offender Guidelines range and thus ineligible for a sentence reduction based on Amendment 706) (emphases in original).

The First and Fourth Circuit Courts similarly have concluded that a career offender who is granted a downward departure under § 4A1.3 to the Crack Cocaine Guidelines range is eligible for a sentence reduction under § 3582(c)(2). See United States v. Cardosa, 606 F.3d 16, 21 (1st Cir.2010); United States v. Munn, 595 F.3d 183 (4th Cir.2010).8 And, like the Second Circuit in McGee (and our Court in Mateo), these Courts agree that a defendant sentenced within the Career Offender Guidelines range would not be eligible for a sentence reduction. See Cardosa, 606 F.3d at 19 (citing United States v. Ayala-Pizarro, 551 F.3d 84, 85 (1st Cir. 2008)); Munn, 595 F.3d at 187 & n. 7.

We agree with the reasoning of these Courts. The Government's contention that Flemming's sentence was "based on" the sentencing range calculated under the Career Offender Guidelines cannot be squared with the ordinary meaning of that phrase. See United States v. Cook, 594 F.3d 883, 888 (D.C.Cir.2010) ("Construed in its ordinary sense, the phrase `based on' refers, for purposes of section 3582(c)(2), to a guideline range that determined the defendant's sentence."); see also Cardosa, 606 F.3d at 20. In applying the § 4A1.3 departure, the District Court did not specify the number of offense levels or criminal history categories by which it was departing. Rather, it simply reverted, without further comment, to the base offense level calculated under the Crack Cocaine Guidelines (24) and the criminal history category that applied to Flemming absent the career offender enhancement (V), and imposed a sentence within that range. Indeed, we have little doubt that had Amendment 706 been in effect when Flemming was sentenced—and, thus, had Flemming's offense level under the Crack Cocaine Guidelines been two levels lower— the District Court would have applied the resulting lower Guidelines range after departing under § 4A1.3.9 To ignore these facts and conclude nonetheless that Flemming was sentenced "based on" the Career Offender Guidelines range would put form over substance. See McGee, 553 F.3d at 228 (reasoning that "a defendant who was, even if by virtue of a departure, sentenced `based on' the crack guidelines [is] eligible for a reduction" under § 3582(c)(2), and noting that "a different reading would lend itself to excessive formalism").

In sum, we conclude that Flemming satisfies the first requirement of § 3582(c)(2), as his sentence was "based on a sentencing range" that has subsequently been lowered by the Sentencing Commission.

B. Does Amendment 706 have the effect of lowering Flemming's "applicable guideline range"?

What initially appears to be a question with a "seemingly commonsense" answer, McGee, 553 F.3d at 229—i.e., whether Flemming's technical status as a career offender, which played no role in the sentence the District Court ultimately imposed, nonetheless makes him ineligible for a sentence reduction-is made far more complicated by the second requirement of § 3582(c)(2), which provides that any sentence reduction must be "consistent with applicable policy statements issued by the Sentencing Commission."10 18 U.S.C. § 3582(c)(2); see also Doe, 564 F.3d at 310.

The policy statement in § 1B1.10 provides that a sentence reduction based on a retroactive amendment is not consistent with that policy statement if the amendment "does not have the effect of lowering the defendant's applicable guideline range."11 U.S.S.G. § 1B1.10(a)(2)(B) (emphasis added). Thus, under the policy statement, Flemming is not eligible for a sentence reduction if his "applicable guideline range" is the sentencing range calculated under the Career Offender Guidelines, as that range is not affected by Amendment 706. However, if Flemming's "applicable guideline range" is the Crack Cocaine Guidelines range—which is affected by Amendment 706—he is eligible for a sentence reduction.

The Government contends that the Guidelines specify that all departures, including the departure authorized under § 4A1.3, result in a sentence outside the "applicable guideline range." Under the Government's theory, it is irrelevant that the District Court rejected the propriety of the career offender enhancement and sentenced Flemming within the Crack Cocaine Guidelines range, because his "applicable guideline range" for purposes of § 1B1.10 remained the Career Offender Guidelines range.

As we explain below, the Government's view, though plausible, is far from compelled by the Guidelines. Rather, after "seiz[ing] every thing from which aid can be derived" to answer this question, Chapman v. United States, 500 U.S. 453, 463, 111 S.Ct. 1919, 114 L.Ed.2d 524 (1991) (internal quotation marks and citation omitted)—i.e., the Guidelines' text, the Sentencing Commission's instructions for applying the Guidelines, and the Commission's applicable Commentary to the Guidelines—we conclude that the edition of the Guidelines used at Flemming's sentencing is ambiguous as to whether the "applicable guideline range" is his pre-§ 4A1.3 departure range (the Career Offender Guidelines range) or his post-§ 4A1.3 departure range (the Crack Cocaine Guidelines range). Under the rule of lenity, we resolve that ambiguity in Flemming's favor.

1. The Application Instructions

The Sentencing Guidelines contain no global definition of the phrase "applicable guideline range," which our Court and other circuit courts have treated as a term of art. As a result, we previously have looked to the Application Instructions for the Guidelines contained in U.S.S.G. § 1B1.1 for guidance in determining the point at which a defendant's "applicable guideline range" is determined. Those Instructions (though they do not use the term "applicable guideline range") instruct sentencing courts to apply the various provisions and chapters of the Guidelines "in a specific order." Doe, 564 F.3d at 311; see also United States v. Johnson, 155 F.3d 682, 684 (3d Cir.1998) (noting that the Application Instructions provide "a sequence of steps for the court to follow in the order in which they appear") (emphasis in original).12 The theory is that, by determining the step of the Application Instructions at which a defendant's "applicable guideline range" has been set, we can identify the adjustments that come after that point (and thus do not affect the "applicable guideline range").

We relied on the Application Instructions in United States v. Doe to determine the "applicable guideline range[s]" for two defendants subject to mandatory minimum sentences of life imprisonment, which exceeded their initial Guidelines ranges of 151-188 and 121-151 months' imprisonment, respectively, calculated under the Crack Cocaine Guidelines. 564 F.3d at 308. The mandatory minimum sentence was not applied to either defendant because each received a downward departure from such a sentence under 18 U.S.C. § 3553(e) and U.S.S.G. § 5K1.1 in exchange for their substantial assistance to the Government. Id. After departing by several levels, the District Court sentenced Jane Doe to 41 months' imprisonment, and John Doe to 84 months' imprisonment. Id.

After Amendment 706 was made retroactive, both defendants filed motions for a reduction of sentence. Id. The Does contended that their "applicable guideline ranges" for purposes of § 1B1.10 were the ranges calculated under the Crack Cocaine Guidelines, as the District Court had relied on those ranges to determine the extent of its departure under § 5K1.1.13 We rejected their argument, concluding that the Does' "applicable guideline range[s]" were the mandatory minimum sentences, rather than their "initial" Guidelines ranges under the Crack Cocaine Guidelines. Id. at 311.

In so holding, we looked to the Application Instructions in § 1B1.1, which culminate in the following four steps:

U.S.S.G. § 1B1.1(f)-(i).

The Does' "initial" Guidelines ranges were determined at step (g), based on the offense levels assigned them under the Crack Cocaine Guidelines in § 2D1.1(c). However, step (h) instructs a sentencing court to determine "the sentencing requirements" for "the particular guideline range" by consulting "parts B through G of Chapter Five," which include application of a mandatory minimum sentence under § 5G1.1. In turn, § 5G1.1(b) provides that, "[w]here a statutorily required minimum sentence is greater than the maximum of the applicable guideline range, the statutorily required minimum sentence shall be the guideline sentence." U.S.S.G. § 5G1.1(b) (emphasis added). In that context, we concluded that (1) the Application Instructions "specif[y] that applying a mandatory sentence pursuant to § 5G1.1(b) [was] the last step in determining the [Does'] . . . Guideline sentence"; and (2) the Does' "initial" advisory ranges under the Crack Cocaine Guidelines had been essentially "`subsumed and replaced'" at step (h) by the mandatory minimum sentences. Id. at 311.

Thus, the implication of our reasoning in Doe is that a defendant's "applicable guideline range," for purposes of § 1B1.10, has been set once a court finishes applying step (h), and that any adjustment applied in step (i)—such as a substantial assistance departure under § 5K1.1—has no effect on the "applicable guideline range," even if a court relied on the defendant's "initial" Guidelines range under the Crack Cocaine Guidelines in determining the extent of such a departure.14

Though Doe did not involve a defendant who received a downward departure under § 4A1.3 to the range calculated under the Crack Cocaine Guidelines,15 our sister circuit courts have similarly relied on the Application Instructions to determine the point at which such a defendant's "applicable guideline range" has been determined. See United States v. Tolliver, 570 F.3d 1062, 1065-66 (8th Cir.2009). In Tolliver, the Eighth Circuit Court held that a career offender who was resentenced within the Crack Cocaine Guidelines range pursuant to a stipulation between the parties was not eligible for a sentence reduction because Amendment 706 did not have the effect of lowering his "applicable guideline range." 570 F.3d at 1064-65. That stipulation followed the defendant's 28 U.S.C. § 2255 motion, in which he argued that he should have received a sentence within the Crack Cocaine Guidelines range (rather than the Career Offender Guidelines range) "as originally contemplated" by the parties in their plea agreement. Id. at 1064. The Government agreed not to oppose the defendant's motion, and stipulated to a sentence within the Crack Cocaine Guidelines range. Id.

Citing our decision in Doe, the Eighth Circuit Court concluded that (1) a defendant's "applicable guideline range" is determined at step (h) of the Application Instructions, and (2) a court determines whether any departures are warranted at step (i), after the "applicable guideline range" has already been set. Id. at 1065-66 (citing Doe, 564 F.3d at 311). Apparently concluding that Tolliver's stipulated sentence reduction was equivalent to a "departure" under the Guidelines (and applied at step (i) of the Application Instructions), the Eighth Circuit determined that his "applicable guideline range" remained the Career Offender Guidelines range and was unaffected by the stipulation.16 Id. at 1066-67. The Court suggested that this conclusion would apply equally with respect to career offenders who received § 4A1.3 departures, and expressed disagreement with the Second Circuit's contrary conclusion in McGee.17 Id. at 1066.

The Sixth and Tenth Circuit Courts have followed the reasoning in Tolliver. See United States v. Pembrook, 609 F.3d 381, 384-86 (6th Cir.2010); United States v. Darton, 595 F.3d 1191, 1194-95 (10th Cir.2010). According to these Circuit Courts, all departures provided for in the Guidelines must be applied after the "applicable guideline range" has been calculated, thus foreclosing the possibility of a departure to the "applicable guideline range." See Pembrook, 609 F.3d at 385-86; Darton, 595 F.3d at 1194.

The Fourth Circuit, however, has concluded that the "plain text" of the Application Instructions compels the opposite conclusion. Munn, 595 F.3d at 193-94.18 As the Munn Court points out, step (i) of the Application Instructions refers only to departures contained in "Part K of Chapter Five," U.S.S.G. § 1B1.1(i), such as a departure for substantial assistance under § 5K1.1. See Munn, 595 F.3d at 193 n. 11. By contrast, step (f) of the Instructions directs a sentencing court to apply "Part A of Chapter Four"—which part includes the departure authorized under § 4A1.3—before a "guideline range" is even mentioned in the Application Instructions (at step (g)). See U.S.S.G. § 1B1.1(g); Munn, 595 F.3d at 192-93. Thus, unlike the substantial assistance departure we considered in Doe, the Application Instructions appear to direct sentencing courts to apply a § 4A1.3 departure at step (f), two steps before a defendant's "applicable guideline range" has been set at step (h).

We agree with the Fourth Circuit that, under one plausible reading of the Application Instructions, sentencing courts are directed to apply § 4A1.3 departures at step (f). Indeed, this seems to flow from common sense. Section 4A1.3 (as construed in Shoupe) and the career offender enhancement in § 4B1.1 both may affect a defendant's offense level and criminal history category—and, as we reasoned in Shoupe, because the "`jump into the career offender category [is] done in one step,'" permitting a departure in both offense level and criminal history category under § 4A1.3 "is reasonable since the district court [is] only undoing the one step." 35 F.3d at 838 (quoting United States v. Bowser, 941 F.2d 1019, 1026 (10th Cir.1991)) (emphasis added). By contrast, a § 5K1.1 substantial assistance departure does not "undo" any Guidelines application decisions. Rather, that departure permits a court to set a sentence outside the "applicable guideline range" based on "a specific factor (the defendant's cooperation) that was not accounted for in the basic guidelines calculation." Roa-Medina, 607 F.3d at 259.

Though we believe the Fourth Circuit's reading of the Application Instructions is plausible, we cannot conclude that the Instructions unambiguously compel this conclusion, as it is not entirely clear that step (f) encompasses a downward departure under § 4A1.3. Section 4A1.3 is a policy statement, and step (i) instructs sentencing courts to consider "any other policy statements. . . in the guidelines that might warrant consideration in imposing sentence." U.S.S.G. § 1B1.1(i). Thus, one could interpret the Application Instructions as directing sentencing courts to apply a § 4A1.3 departure at the final step (step (i)), after the "applicable guideline range" has already been calculated. See Pembrook, 609 F.3d at 387; Munn, 595 F.3d at 196-97 (Duncan, J., dissenting).

This competing interpretation of § 1B1.1 would be more compelling if step (i) of the Application Instructions encompassed the full universe of "other policy statements" in the Guidelines "that might warrant consideration in imposing sentence." U.S.S.G. § 1B1.1(i). But that is not the case. For example, step (h) of the Instructions specifically directs sentencing courts to apply "parts B through G of Chapter Five" of the Guidelines, which parts include several policy statements. See U.S.S.G. § 5B1.3(c), (d), (e); § 5D1.3(c), (d), (e); § 5E1.5; § 5F1.7; § 5G1.3(c). Accordingly, step (i)'s reference to "other policy statements" does not foreclose the possibility that step (f) encompasses the downward departure authorized by the policy statement in § 4A1.3.

In the end, we conclude that the Application Instructions are ambiguous as to the step at which a sentencing court must apply a § 4A1.3 downward departure. The Instructions do not clearly require the departure to be applied after the "applicable guideline range" is calculated (as the Sixth, Eighth, and Tenth Circuits have concluded), nor do they clearly require the departure to be applied before the "applicable guideline range" is calculated (as the Fourth Circuit has concluded).19 Accordingly, the Application Instructions do not, by themselves, answer whether Flemming's "applicable guideline range" is that set by Crack Cocaine Guidelines or the Career Offender Guidelines.

2. The 2003 Amendment to the § 1B1.1 Commentary

Our conclusion in Doe that the defendants' "applicable guideline range[s]" were the mandatory minimum sentences was based not only on our reading of the Application Instructions, but also on the Sentencing Commission's commentary to the policy statement in § 1B1.10, which strongly suggests that a sentence reduction is not authorized where a defendant is subject to a mandatory minimum sentence:

Doe, 564 F.3d at 312 (quoting U.S.S.G. § 1B1.10 cmt. 1(A)) (first omission, alteration, and emphasis in original).

In concluding that a defendant in Flemming's position is not eligible for a sentence reduction, the Sixth, Eighth, and Tenth Circuits relied heavily on a 2003 amendment to the Guidelines, Amendment 651, that appears similarly to suggest that a § 4A1.3 downward departure has no effect on a defendant's "applicable guideline range." In particular, Amendment 651 added, for the first time, an explicit definition of the term "departure" to the Commentary to U.S.S.G. § 1B1.1:

U.S.S.G. § 1B1.1 cmt. n. 1(E) (2003).

This new definition of "departure" appears to indicate that a § 4A1.3 downward departure is a departure from, rather than to, the "applicable guideline range," as the commentary now states that a § 4A1.3 downward departure is applied "in order to effect a sentence outside the applicable guideline range." Id. (emphasis added). The Sixth, Eighth, and Tenth Circuits seized on this definition to conclude that a § 4A1.3 departure has no effect on the "applicable guideline range" for a career offender.20 See Tolliver, 570 F.3d at 1066 (citing the definition of "departure" added in 2003 and concluding that the Application Instructions "effectively define all departures to be outside the `applicable guideline range'"); see also Pembrook, 609 F.3d at 385-86; Darton, 595 F.3d at 1194.

However, we agree with the Fourth Circuit that another Guidelines provision not considered by the Sixth, Eighth, or Tenth Circuits— § 1B1.11—precludes us from considering the 2003 definition of "departure" in determining Flemming's eligibility for a sentence reduction. Section 1B1.11 provides that

U.S.S.G. § 1B1.11(b)(2) (emphasis added). Flemming was sentenced under the 2001 edition of the Guidelines. Accordingly, we must consider Amendment 651 if it is a "clarifying" amendment. Id. However, if the 2003 definition makes a "substantive change[]" to the Guidelines, we may not consider it. Id.; see also United States v. Marmolejos, 140 F.3d 488, 491 (3d Cir. 1998) (courts must determine whether an amendment "clarified the existing commentary in the Guidelines or substantively changed its meaning") (emphases in original).

"Generally, if [an] amended guideline and commentary overrule[] a prior judicial construction of the guidelines, it is substantive; if it confirms our prior reading of the guidelines and does not disturb prior precedent, it is clarifying." United States v. Diaz, 245 F.3d 294, 303 (3d Cir. 2001). We recently held that Amendment 651 overrides our holding in Shoupe because, as a result of the new definition of "departure," § 4A1.3 now authorizes a departure in criminal history category only.21 See Grier, 585 F.3d at 143. Thus, because the 2003 definition worked a substantive change in the law, i.e., it overruled our interpretation of § 4A1.3 in Shoupe, Amendment 651 is a substantive amendment that we may not consider in determining whether Flemming is eligible for a sentence reduction.22 Accord Munn, 595 F.3d at 194 (noting that Amendment 651 "did more than simply clarify the Guidelines: [i]t effected a significant change in the law of this and nearly every other circuit"). Importantly, both the Government and Flemming agreed at oral argument that the 2003 definition of "departure" is a substantive amendment for purposes of U.S.S.G. § 1B1.11.

In that light, we find the Sixth, Eighth, and Tenth Circuit's reasoning unpersuasive. In Tolliver, the Eighth Circuit did not explain why it was authorized to consider the 2003 definition of "departure," given that the defendant in that case—like Flemming and the defendants in Munn, McGee, and Cardosa—was sentenced under a pre-2003 edition of the Sentencing Guidelines.23 See Munn, 595 F.3d at 193 n. 12. The defendant in the Sixth Circuit's Pembrook decision was also sentenced under a pre-2003 version of the Sentencing Guidelines. See Pembrook, 609 F.3d at 382. But though the Sixth Circuit Court acknowledged the Fourth Circuit Court's conclusion that the 2003 definition of "departure" could not be considered in those circumstances, it (the Sixth Circuit) nonetheless emphasized, without providing any basis for rejecting the Munn Court's conclusion, that § 4A1.3 "[i]n its present form" makes clear that a downward departure under that provision is one from the "applicable guideline range."24 Id. at 385-87.

3. The 2001 Version of § 4A1.3

Not only are we precluded from considering the 2003 definition of "departure," the 2001 version of § 4A1.3, which lacked any such definition, appears to differ materially from the 2003 version with respect to what constitutes a defendant's "applicable guideline range" following a departure under that provision. The 2001 version of § 4A1.3 provides in relevant part:

U.S.S.G. § 4A1.3 (2001) (emphasis added). The majority in Pembrook and the dissenter in Munn concluded that § 4A1.3's reference to "imposing a sentence departing from the otherwise applicable guideline range" conveys the same meaning as the 2003 definition of "departure," i.e., that a § 4A1.3 departure does not affect a defendant's "applicable guideline range." Pembrook, 609 F.3d at 385-86; Munn, 595 F.3d at 196 (Duncan, J., dissenting).

We disagree. To us, this language facially supports the conclusion that a defendant's "applicable guideline range" is set after a downward departure under § 4A1.3. That is, under the plain meaning of the word "otherwise," § 4A1.3 appears to contemplate that a new "applicable guideline range" results from a downward departure pursuant to that provision, replacing the "otherwise applicable guideline range." See United States v. Mobley, 956 F.2d 450, 452 (3d Cir.1992) ("We construe terms of the Guidelines according to their plain meaning."); see also Webster's Third New Int'l Dictionary 1598 (1971) (the definitions of the adverb "otherwise" include "in a different way or manner," "in different circumstances," and "under other conditions"). Were that not the case, the word "otherwise" would be meaningless, as the contrary interpretation equally would make sense if § 4A1.3 provided that a court may depart "from the applicable guideline range." Cf. Acceptance Ins. Co. v. Sloan, 263 F.3d 278, 283 (3d Cir.2001) ("[It is an] axiom of statutory construction that whenever possible each word in a statutory provision is to be given meaning and not to be treated as surplusage." (internal quotation marks and citation omitted)).

This reading of the 2001 version of § 4A1.3 is bolstered when we compare its text to provisions in Chapter Five of the Guidelines. The provisions in Parts H through K of Chapter Five—which courts must consider at the final step of the Application Instructions (step (i))—set forth the circumstances in which a court may impose a sentence "outside" or "below" "the applicable guideline range." See U.S.S.G. §§ 5H1.1-6, 11-12, 5K2.0, 5K2.12, 5K2.13, 5K2.16, 5K2.20 (2001). By contrast, the 2001 version of § 4A1.3 refers to a departure "from the otherwise applicable guideline range," presumably contemplating that the range that results from a § 4A1.3 departure "`subsumes and displaces the otherwise applicable guideline range,'" Doe, 564 F.3d at 311 (quoting United States v. Cordero, 313 F.3d 161, 162 (3d Cir.2002)) (emphasis added), and thus becomes "the applicable guideline range." Applying this reading to Flemming's situation, the Career Offender Guidelines range was his "otherwise applicable guideline range," but was replaced by the Crack Cocaine Guidelines range as a result of the § 4A1.3 downward departure.

This interpretation has some surface appeal. We note, however, that numerous variations of the term "guideline range," of which "otherwise applicable guideline range" appears to be one, appear in other departure provisions located in Part K of Chapter Five. These variations include "the range established by the applicable guidelines," U.S.S.G. § 5K2.0 (2001), "the authorized guideline range," id. §§ 5K2.1-7, "the guideline range," id. §§ 5K2.8-10, 5K2.14, and, more simply, "the guidelines," id. § 5K1.1. Other provisions in Part K make no reference to a "guideline range" or the "applicable guideline range," instead simply noting the circumstances that might warrant an upward or downward "departure," id. §§ 5K2.17-19, or "a reduced sentence," id. § 5K2.11. It is undisputed that these departures are all applied at the final step of the Application Instructions and, when applied, result in a sentence outside the "applicable guideline range." However, that these provisions do not employ the term "the applicable guideline range" in describing such a departure may suggest that, though the Guidelines do not use consistent terminology to describe the range from which a sentencing court may depart in the final step, this fact does not necessarily suggest that any particular variation was intended to be dispositive, or even have significance, with regard to a defendant's eligibility for a sentence reduction under § 1B1.10.

In sum, we conclude that the text of the 2001 version of § 4A1.3, like the Application Instructions, fails to answer unambiguously whether Flemming's "applicable guideline range" is his pre- or post- § 4A1.3 departure range.

4. The Rule of Lenity

The rule of lenity provides that "when ambiguity in a criminal statute cannot be clarified by either its legislative history or inferences drawn from the overall statutory scheme, the ambiguity is resolved in favor of the defendant." United States v. Pollen, 978 F.2d 78, 85 (3d Cir. 1992). The rule covers criminal prohibitions as well as penalties, see Mobley, 956 F.2d at 452 (citing Bifulco v. United States, 447 U.S. 381, 100 S.Ct. 2247, 65 L.Ed.2d 205 (1980)), and applies to the Sentencing Guidelines. See United States v. Fenton, 309 F.3d 825, 828 n. 3 (3d Cir.2002) ("[W]here . . . the [Sentencing] Guidelines do not clearly call for enhancement, the rule of lenity should prevent the application of a significantly increased sentence."); see also United States v. Simpson, 319 F.3d 81, 87 (2d Cir.2002) (noting that "the rule of lenity is generally applicable to the Sentencing Guidelines").25

Application of the rule of lenity requires more than a difficult interpretative question. Rather, "[t]o invoke the rule, we must conclude that there is a grievous ambiguity or uncertainty in the statute." Muscarello v. United States, 524 U.S. 125, 138-39, 118 S.Ct. 1911, 141 L.Ed.2d 111 (1998) (internal quotation marks and citation omitted) (emphasis added); see also Chapman, 500 U.S. at 463, 111 S.Ct. 1919. As discussed, we believe the Guidelines are "grievous[ly] ambiguous [and] uncertain[]" as to whether Flemming's "applicable guideline range" is his pre- or post-§ 4A1.3 departure range. Muscarello, 524 U.S. at 139, 118 S.Ct. 1911. Because both interpretations are plausible, we apply the rule of lenity and resolve the ambiguity in Flemming's favor. See United States v. Oetken, 241 F.3d 1057, 1060 (8th Cir.2001) ("Where there are two plausible readings of a guideline provision, we apply the rule of lenity and give the defendant the benefit of the reading that results in the shorter sentence."); see also Bifulco v. United States, 447 U.S. 381, 387, 100 S.Ct. 2247, 65 L.Ed.2d 205 (1980) ("[T]he touchstone of the rule of lenity is statutory ambiguity." (internal quotation marks and citation omitted)). We join the Second Circuit in this regard. See McGee, 553 F.3d at 229 (applying the rule of lenity while acknowledging that "U.S.S.G. § 1B1.10 can be read to permit a reduced sentence only where the defendant's pre-departure sentencing range is found within the crack cocaine guidelines") (emphasis in original); see also Pembrook, 609 F.3d at 391-92 (Rogers, J., dissenting) (arguing that application of the rule of lenity is appropriate in these circumstances, as "there is no compelling reason to read the phrase `applicable guideline range' to exclude the Guideline range that the district court actually applied to [the defendant]").

Though the 2003 definition of "departure" may resolve this ambiguity, Flemming was sentenced under the 2001 edition of the Sentencing Guidelines, and Amendment 651 is a substantive amendment that we may not consider. In any event, the Sentencing Commission's decision in 2003 to add a uniform definition of "departure" to the Guidelines—which includes a definition specific to § 4A1.3—only bolsters our conclusion that, at least prior to 2003, what constituted the "applicable guideline range" for a career offender granted a downward departure under § 4A1.3, and sentenced based on the Crack Cocaine Guidelines range, was, at best, quite unclear.26

We reach this result aware not only of the general policies behind retroactive amendments to the Sentencing Guidelines, but also the particular policy underlying Amendment 706. Cf. Moskal v. United States, 498 U.S. 103, 108, 111 S.Ct. 461, 112 L.Ed.2d 449 (1990) (noting that courts may consult the "motivating policies of the statute" in determining whether application of the rule of lenity is warranted (internal quotation marks and citation omitted)). The Sentencing Reform Act of 1984, 98 Stat. 1987, charged the Sentencing Commission with periodically reviewing and revising the Guidelines, and, where warranted, authorized the Commission to reduce retroactively the offense levels for certain crimes. See 28 U.S.C. §§ 994(o), 994(u); see also 18 U.S.C. § 3582(c). When the Commission first promulgated the Guidelines in 1987, it adopted offense levels for cocaine offenses that "treated every gram of crack cocaine as the equivalent of 100 grams of powder cocaine." Kimbrough v. United States, 552 U.S. 85, 96, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007). The Commission later attempted to alleviate this disparity and, after Congress repeatedly failed to act, id. at 99, 128 S.Ct. 558, the Commission adopted Amendment 706 to address what it believed were "urgent and compelling" problems with the 100-to-1 ratio. U.S.S.G. app. C, amend. 706, Reason for Amendment (2007).

There is no dispute that Flemming was burdened by the very crack/powder cocaine disparity that the Sentencing Commission sought to remedy by promulgating Amendment 706 and making it retroactive. As noted, had Amendment 706 been in force when Flemming was sentenced, we have little doubt the District Court would have set a sentence within that range. In these circumstances, we believe that rendering a defendant ineligible for a sentence reduction—simply because he technically qualified as a career offender, and despite a District Court's reasoned judgment that such a classification was inappropriate because it overstated the seriousness of his criminal history (in Flemming's case, because his two predicate offenses resulted in modest prison sentences and were committed when he was 18 and 20-years old)—is antithetical to the policy concerns that motivated Amendment 706. See Cardosa, 606 F.3d at 21; Munn, 595 F.3d at 195; McGee, 553 F.3d at 229-30.

* * * * * *

Application of the rule of lenity is called for only in rare cases, and thus we stress the narrowness of our holding. We conclude that, under a pre-2003 edition of the Sentencing Guidelines, a career offender who is granted a § 4A1.3 downward departure to the Crack Cocaine Guidelines range is eligible for a sentence reduction under 18 U.S.C. § 3582(c)(2). Accordingly, we vacate the District Court's order and remand this case for the Court to exercise its discretion to determine whether, and to what extent, a reduction in Flemming's sentence is warranted.