Appellant, Kenneth L. Kenley (hereinafter "Kenley"), was convicted in separate trials of capital murder and related noncapital offenses based on a spree of robberies that occurred on January 3 and 4, 1984. A summary of the crime facts appears in the opinion of the Missouri Supreme Court. State v. Kenley, 693 S.W.2d 79, 80-81 (Mo.1985) (en banc). In sum, Kenley committed one murder, shot two bystanders, abducted two bystanders, stole two cars, committed several assaults, and robbed four businesses in just a few hours. Kenley was represented by appointed counsel. The two separate cases were based on essentially the same evidence, which included seven eyewitnesses to the murder and twenty-six total witnesses corroborating various aspects of Kenley's identity, his vehicles and his weapon. At the capital trial the jury found two aggravating circumstances, murder for value and the knowing creation of risk of death in a public place, and imposed the death sentence.

The capital appeal unsuccessfully alleged errors related to evidentiary issues, jury procedures and violation of the eighth amendment. Kenley, 693 S.W.2d at 81-83. The noncapital appeal unsuccessfully alleged errors related to evidentiary issues and ineffective assistance of counsel. State v. Kenley, 701 S.W.2d 185 (Mo.App.1985). Kenley then sought post-conviction relief from both the capital and noncapital sentences, received a hearing, but was denied relief by the Missouri Circuit Court (hereinafter "circuit court"). The claims were then reviewed by the Missouri Court of Appeals (hereinafter "appeals court") which affirmed the judgment of the circuit court. Kenley v. State, 759 S.W.2d 340 (Mo.App.1988).

This appeal arises from the federal district court's (hereinafter "district court") denial of Kenley's petition for writ of habeas corpus. 28 U.S.C. Sec. 2254 (1990). The petition raises thirty-five grounds for relief as to the capital conviction and a dozen grounds for relief as to the noncapital convictions (primarily claims of ineffective assistance of counsel).

Kenley argues, with respect to both the capital and noncapital claims, that the district court erred by: (1) finding no ineffective assistance of counsel during the guilt or sentencing phase of either trial, (2) finding many of Kenley's claims for relief procedurally barred, and (3) applying the presumption of correctness to state court findings too broadly. Because we believe Kenley received ineffective assistance of counsel during the sentencing phase of his capital trial, we reverse the judgment of the district court and remand this case to the district court with instructions that the State of Missouri resentence Kenley.

The facts that follow represent a composite of the consistent findings made by the Missouri courts and the district court, unless otherwise stated. Appointed counsel was a new county public defender just three months out of law school with no trial experience. The trial occurred six months after appointment when counsel had one misdemeanor robbery trial under his belt. Counsel's time card from Butler County, Missouri indicates he spent roughly fifty hours in pre-trial preparation for Kenley's capital trial.

Counsel sought and received little outside assistance in preparing his case. Kenley's former social worker, Mrs. Crownover, initiated contact with counsel offering assistance, and counsel responded with a request for her records. The parties had no further contact. A letter from the Office of State Public Defender was also sent to counsel suggesting he associate with co-counsel, and offering case assistance in rural counties (such as Butler County) from more experienced public defenders elsewhere. Counsel did not respond to this letter but did testify in the state post-conviction proceeding that he spoke with a former classmate and another attorney working as public defenders in St. Louis about the case. His former classmate had only misdemeanor experience.

Crownover's letter referred counsel to a Dr. Manion who had previously examined and treated Kenley in 1975. Counsel never contacted Manion. After receiving Crownover's records, counsel requested a psychiatric examination which was performed by Dr. Parwatikar in 1984 at Fulton State Hospital. Counsel testified that he discussed the Parwatikar report with a non-examining physician, Dr. Wiscowski, for six hours. This is not corroborated by his time card. Counsel testified he then decided not to introduce Kenley's background as mitigating evidence because the information was too old and insubstantial, Parwatikar's report was conclusively non-mitigating, and it was possible additional aggravating information would also be introduced. Kenley, 759 S.W.2d at 345. Counsel apparently did not seek or review any other records of Kenley's background.

Each court before us reviewed the Crownover, Manion and Parwatikar information reviewed by counsel. Representative of their findings is this summary by the appeals court:

... Kenley had been aggressive, hostile and went into uncontrollable rages and had shown improvement only while on medication in an institution; that Kenley grew up in a non-nurturing environment and had a violent and angry relationship with his father; that all attempts at control, help and treatment of him had proven futile; that Dr. Manion expressed the opinion that "treatment will facilitate adjustment, further adjustment"; and that the latest documentation by either witness was at least 6 years old.

Kenley, 759 S.W.2d at 347.

Prior to the capital trial, counsel waived an automatic change of venue despite the wide publicity Kenley's crimes had received. During voir dire, counsel performed little inquiry into the attitudes of the jurors toward capital punishment and into jurors' ability to consider a range of possible punishments. Counsel waived opening statement and only asked thirty-six cross-examination questions of eight out of twenty-six witnesses. Counsel did not challenge somewhat inaccurate descriptions of Kenley and his vehicle, did not challenge ambiguities and inconsistencies in witness statements, and did not object to a surprise pathologist's report. Counsel also did not object to the admission of cumulative and potentially prejudicial testimonial evidence regarding Kenley's identity, the victim's family, and other weapons unrelated to the crime. Counsel made a short, unremarkable closing argument and submitted no jury instructions. Capital murder, second degree murder and manslaughter instructions were given by the court.

At the capital sentencing, counsel submitted no mitigating evidence although Kenley's age was stipulated and the prosecution introduced Kenley's single prior conviction. Counsel unwittingly made reference to Kenley's failure to testify and did not object to evidence regarding the deterrent effect of the death penalty introduced by the prosecution. Many similar events occurred (or failed to occur) at Kenley's noncapital trial. In addition, counsel made no closing argument and did not move for severance of charges (robbery, kidnapping, assault and theft). By the close of the noncapital trial, counsel had accepted a job with the prosecuting attorney's office.

The district court, applying the presumption of correctness, adopted the state court findings regarding the Crownover, Manion and Parwatikar evidence. It characterized the evidence as "scant" mitigating evidence. Mem.Op. at 40. The district court ruled that Kenley had procedurally defaulted on at least some part of twenty-five of his thirty-five claims for relief from the capital conviction and on two of his twelve claims for relief from the noncapital convictions. The court concluded cause and prejudice had not been demonstrated on any barred claims. Id. at 4. The district court determined the remaining claims were without merit after deferentially reviewing the state court fact findings and legal conclusions.

Likewise, this court undertakes a deferential review when considering the factual findings of the state and district courts. As we have consistently said in recent decisions, we apply a presumption of correctness to state court findings of fact, we review claims of ineffective assistance of counsel de novo as a mixed question of fact and law, and we only review those claims which were fairly presented first to the state courts and the federal district court, unless cause and prejudice or miscarriage of justice are shown sufficient to excuse default. See, e.g., Bolder v. Armontrout, 921 F.2d 1359, 1361 (8th Cir.1990), reh'g denied, 928 F.2d 806 (8th Cir.1991); Laws v. Armontrout, 863 F.2d 1377, 1381-82 (8th Cir.1988) (en banc), cert. denied, 490 U.S. 1040, 109 S. Ct. 1944, 104 L. Ed. 2d 415 (1989).

The petition for habeas relief contained several different claims of ineffective assistance of counsel and other claims based on allegedly inadmissible evidence, incorrect jury procedures and instructions, and violation of the eighth amendment. Having reviewed the relevant opinions of the respective courts and the excerpts from state court filings contained in the record, we affirm the district court's determination of procedural bar with regard to twelve of Kenley's claims.1 These claims were not adequately presented to the state courts, and Kenley did not meet his burden of showing cause and prejudice or miscarriage of justice. See Murray v. Carrier, 477 U.S. 478, 106 S. Ct. 2639, 91 L. Ed. 2d 397 (1986); Wainwright v. Sykes, 433 U.S. 72, 97 S. Ct. 2497, 53 L. Ed. 2d 594 (1977).

As for the other claims determined to be in default by the district court, we find probative evidence in the record and in the text of the district court's memorandum opinion that suggests these issues may have been sufficiently presented to avoid default.2 However, notwithstanding this possibility, we conclude that all but two of these claims were either thoroughly and adequately addressed by the state courts, or were insufficient, separately and collectively, to justify habeas relief in light of the overwhelming evidence of guilt.3 See Long v. Iowa, 920 F.2d 4, 6 n. 2 (8th Cir.1990) (at least where relief is denied, court may bypass procedural bar and address the merits if it is the simpler issue). With respect to the claims denied by the district court on the merits,4 we conclude that as to all but two of these claims the district court did not err in applying the presumption of correctness to state court fact findings, and we do not disagree with its conclusions of law. The denied claims are simply insufficient to justify relief in light of the overwhelming evidence of guilt or are otherwise meritless.

Of the claims the district court denied on the merits, this leaves for our consideration capital claims 2 and 33. Claim 2 states that "trial counsel was ineffective at the punishment stage of petitioner's capital murder trial in that he failed to present the medical, psychological and psychiatric testimony and reports" of Crownover and Manion. Claim 33 asserts ineffectiveness for failing to present evidence of prior Farmington State Hospital admissions. We also consider capital claims 1 and 27 because excerpts from the state court filings quoted in the record and the text of the information considered by the state courts persuade us that these claims were adequately presented to the state courts to avoid procedural default. Furthermore, though stating the two claims were barred, the district court considered their merits to a great extent because of overlap with claims it considered on the merits.

Claims 1 and 27 charge that counsel failed to present available mitigating evidence and failed to contact or call family members as witnesses during sentencing. In Kenley's pro se post-conviction relief petition, claim [b], Kenley states "[c]ounsel failed to present available mitigating evidence to the jury during the penalty phase portion of the trial," and then follows with nonexclusive examples of specific witnesses and reports, including "such witnesses and records documenting a mental condition of movant which would have been admitted as mitigating circumstances both as to guilty [sic] and punishment." Mem.Op. at 9. In claim [c] Kenley states "[c]ounsel failed to properly investigate movant's case," and refers to specific failures to pursue referenced leads contained in his social history. Id. In Kenley's brief on appeal from denial of post-conviction relief, claim , he states "counsel failed ... to investigate and present such evidence in that counsel had access to lay and psychiatric evidence concerning Kenley's background, turbulent and abused family history, and psychiatric record [sic] including evidence that his condition could yet be improved (emphasis added)." Id. at 11-12.

The circuit court characterized Kenley's claim as "failure to produce mitigating evidence at the sentencing phase of the trial" and later as a general failure to produce mitigating evidence. Circuit Court Opinion at 1, 3. The appeals court characterized Kenley's point of error broadly as a failure to investigate for and present mitigating psychiatric evidence. Kenley, 759 S.W.2d at 344. The district court similarly characterized the issue as counsel's decision to not further investigate and present mitigating psychiatric evidence. Mem.Op. at 38. The district court also mentioned counsel's investigation into "other information potentially relevant to mitigating circumstances," though it gave no details and made no findings regarding that investigation. Id. at 38.

The principal evidence which Kenley seeks to have us consider relating to the barred claims is unreviewed medical reports of Dr. Richards and from the military, and family testimony regarding Kenley's background and recent observations of his behavior and condition corroborative of his diagnosed psychiatric and substance abuse problems.

We have not applied an unreasonable standard when deciding whether habeas claims have been adequately presented to the state court. The same factual arguments and legal theories should be present in both the state and federal claims. Tippitt v. Lockhart, 903 F.2d 552, 554 (8th Cir.), cert. denied, --- U.S. ----, 111 S. Ct. 301, 112 L. Ed. 2d 254 (1990). This means that the federal claim should not present significant additional facts such that the claim was not fairly presented to the state court. Stranghoener v. Black, 720 F.2d 1005, 1007-08 (8th Cir.1983). We may review claims as long as they are closely related to claims presented to the lower courts, Guinan v. Armontrout, 909 F.2d 1224, 1226-27 (8th Cir.1990), cert. denied, --- U.S. ----, 111 S. Ct. 800, 112 L. Ed. 2d 861 (1991), but we have required at least an "arguable factual commonality." See Bolder, 921 F.2d at 1364 n. 9 (quoting Laws, 863 F.2d at 1387 n. 10).

As stated, the failure to investigate and present expert and lay mitigating evidence of Kenley's background and medical problems was raised by the filings in state court. The state courts discussed counsel's failure to present mitigating evidence and his decision not to investigate further. The district court considered these same issues. Mem.Op. at 38. The transcript of the post-conviction relief hearing contains admissions by Kenley and counsel that they discussed the topic of family witnesses. Dr. Parwatikar's report states that he contacted family witnesses. The records and reports considered by the courts documented the existence of Dr. Richards's report, Kenley's military service, names of family members familiar with Kenley's problems, and Kenley's substance abuse problems. Therefore, claims 1 and 27 were presented to the state courts and appear to be substantially based on the same facts and evidence as the two other claims before us. Applying the stated standards to these facts, we conclude that claims 1 and 27, as they relate to the Dr. Richards and military reports, and to the family testimonial evidence we mentioned, were properly presented to the state courts and our review is not barred.

Based on the preceding discussion, the precise issue before us is the question whether Kenley received ineffective assistance of counsel during the sentencing phase of his capital trial due to counsel's failure to present available family and expert mitigating evidence of his medical, psychological and psychiatric history. The primary evidence that Kenley relies upon for reversal is his social history, his pre-examination report, reports by Drs. Manion and Richards, his military record and family witness testimony.

Ineffectiveness during the guilt phase, though briefed by Kenley, is not now at issue. As we stated, many of his claims raising this issue were thoroughly and adequately addressed by the state courts or were otherwise meritless. Some were defaulted and others were redundant. As to the guilt phase ineffectiveness claims reviewed by the district court, we do not opine as to whether counsel's performance failed to meet the minimum constitutional standard because we find no prejudice given the overwhelming evidence of guilt. This included the in-court testimony of seven eyewitnesses to the murder and the corroborating testimony of twenty-six total witnesses. See Strickland v. Washington, 466 U.S. 668, 697, 104 S. Ct. 2052, 2069, 80 L. Ed. 2d 674 (1984) (performance need not be evaluated if there is no prejudice).

To grant relief based on a claim of ineffective assistance of counsel, we must find (1) counsel's performance was constitutionally ineffective (performance test) and (2) the ineffective performance resulted in prejudice (prejudice test). Id. at 687, 104 S.Ct. at 2064. With the performance test, we start by presuming that counsel provided effective assistance, and we do not use hindsight to question counsel's performance. Id. at 689-90, 104 S.Ct. at 2065-66. Counsel's performance is judged based on an objective standard of what would be considered reasonably effective performance in the circumstances of a particular criminal proceeding. Id. at 687-88, 104 S.Ct. at 2064-65.

With the prejudice test, we evaluate the outcome of the proceeding in question. The ineffective performance must create a reasonable probability that the result of the proceeding is unreliable; the ineffectiveness must undermine our confidence in the proceeding. Id. at 693-94, 104 S.Ct. at 2067-68. This means more than just a conceivable effect on the outcome but something less than the oft used "more likely than not" or preponderance standards. Id. In the context of a death sentence, we must determine the question whether, but for counsel's deficient performance, there is a reasonable probability the jury would have concluded the balance of the aggravating and mitigating circumstances did not warrant death. Id. at 695, 104 S.Ct. at 2069.

The Supreme Court requires that counsel make a reasonable investigation in the preparation of a case or make a reasonable decision not to conduct a particular investigation. Id. at 691, 104 S.Ct. at 2066. Counsel's "strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation." Id. at 690-91, 104 S.Ct. at 2066. We have also considered these principles in our circuit and have held that counsel must exercise reasonable diligence to produce exculpatory evidence and strategy resulting from lack of diligence in preparation and investigation is not protected by the presumption in favor of counsel. Eldridge v. Atkins, 665 F.2d 228, 235-37 (8th Cir.1981), cert. denied, 456 U.S. 910, 102 S. Ct. 1760, 72 L. Ed. 2d 168 (1982).5 Failing to interview witnesses or discover mitigating evidence relates to trial preparation and not trial strategy. Chambers v. Armontrout, 907 F.2d 825, 828 (8th Cir.) (en banc), cert. denied, --- U.S. ----, 111 S. Ct. 369, 112 L. Ed. 2d 331 (1990). Counsel's performance may be found ineffective if s/he performs little or no investigation. See, e.g., id. at 829-31 (counsel's failure to investigate enough to find substantial, available mitigating evidence is ineffective assistance); Thomas v. Lockhart, 738 F.2d 304, 308 (8th Cir.1984) (counsel's reliance on medical report and interviews with defendant only was an inadequate investigation); Pickens v. Lockhart, 714 F.2d 1455, 1467 (8th Cir.1983) (abdication of investigative duty is ineffective assistance of counsel). We have stated that failing to present mitigating evidence may be ineffective assistance if, due to inadequate trial preparation and investigation, "counsel has through neglect failed to discover such evidence...." Laws, 863 F.2d at 1385.

Kenley contends that the Crownover, Manion and Parwatikar evidence, and his discussions with counsel, should have led counsel to further investigate for mitigating evidence. We must evaluate the reasonableness of counsel's decision not to investigate further in light of what he knew or should have known.

The state primarily focussed on counsel's decision not to present any mitigating evidence rather than the decision not to investigate further, but it appears the same reasoning was relied upon for both analyses. In reaching the legal conclusion that counsel's decision was reasonable, the state courts relied on counsel's conclusions that available mitigating evidence was too insubstantial and too old, that Dr. Parwatikar's findings were conclusively non-mitigating, and that additional aggravating information would exceed mitigating information. The first two contentions we address now; the third we save for our prejudice discussion. The district court concluded, without discussion, that counsel had conducted a thorough investigation and his decision not to investigate further was reasonable in light of the information he had. Mem.Op. at 41. We have some trouble with these unsupported conclusions.

The fact findings made by the state courts regarding the Crownover, Manion and Parwatikar evidence that were used to support their legal conclusion were fairly similar. The circuit court which held Kenley's state post-conviction proceeding concluded:

The sum total of Mrs. Crownover's testimony would be that as a young boy between the ages of 11 years and 15 years, the defendant had been aggressive, hostile and not functioning in the classroom and as a result of such conduct, he had been expelled from school. The testimony of Dr. Manion would have been that as a teenager the defendant would go into a state of uncontrollable rage, that he was extremely defensive, difficult to work with, not communicative, expressed negative and hostile feelings toward his family, especially his father, that he had only really shown any improvement while on medication in an institutional situation.

Circuit Court Opinion at 3-4. The circuit court also concluded that "[n]either Dr. Manion's nor Mrs. Crownover's testimony would conclusively show that the defendant was suffering from any mental disease or defect." Id. at 4. The appeals court reached a similar conclusion. Kenley, 759 S.W.2d at 347.

We think these fact findings are unfairly slanted against Kenley, are incomplete and are only partially supported by the record. We are therefore unable to deferentially and exclusively rely on them. In an attempt to justify counsel's decision, the state and district courts have trivialized the information contained and referred to in the Crownover and Manion records, characterizing it as "scant" mitigating evidence, and have further overstated the conclusiveness of Dr. Parwatikar's report. Most of the mitigating information has been omitted from their fact findings. By way of supplement we offer additional facts adduced from the materials which were available to counsel and the state courts to better reflect the true picture of mitigating and aggravating evidence reviewed by counsel.

Kenley was first treated at the Poplar Bluff, Missouri diagnostic center in 1971. He was next treated at the outpatient clinic of Farmington State Hospital in 1973, then again in 1975. The admissions appear to have been related to behavioral problems at home and in school. Kenley was treated again at the Poplar Bluff diagnostic center in 1975 after he experienced fainting spells. From there he was sent to the Lutheran Medical Center where many of his medical problems were first documented. The records contain information on suicide attempts, abnormal tracing in Kenley's EEG which was considered "not conclusive of convulsive disorder," drug use, anxiety and hostility. The records contain a reference to the fact Kenley was sleeping with a gun under his bed for fear of his father. His problem was diagnosed as a possible seizure disorder. That same year Kenley was first treated by Dr. Manion. In 1977 Kenley was treated by Dr. Richards at Farmington State Hospital and was also discharged from the military due to a suicide attempt. Names of family members appeared at various points in the reports and records with respect to incidents and interviews. This information was all referred to in one or more of Crownover's records, the pre-examination questionnaire, and Parwatikar's report.

Before Dr. Parwatikar's examination of Kenley, Crownover completed a pre-examination questionnaire. In that questionnaire and in her social history she documented Kenley's troubled home life and childhood, including the divorce of his parents, environmental instability, a father with psychiatric problems who threatened Kenley on many occasions, and aggressive, sometimes violent behavior by Kenley. She also documented Kenley's alcohol problem, the prescription drug treatments, and that Kenley served in the military.

Dr. Manion's records document Kenley's report of his father's threats. They also document that Kenley's father had a chronic mental illness and was a continuing psychiatric out-patient. Dr. Manion concluded Kenley had an unsocialized behavior disorder and prescribed anti-depressants used for mental patients to try to normalize his behavior. The medication resulted in some improvement in Kenley's behavior, and Dr. Manion opined that with continued treatment Kenley "would be improved." Kenley's EEG was considered abnormal by Manion but was inconclusive of convulsive disorder.

After considering interviews with Kenley and other limited background information, Dr. Parwatikar performed intelligence, brain function and personality tests on Kenley. Parwatikar concluded Kenley had an anti-social personality disorder (based on personality test results), but was competent to stand trial and was not suffering from an exculpatory mental disease or defect. Dr. Parwatikar reached no diagnosis regarding the results of Kenley's intelligence and organic brain functioning tests, although his report notes some tests were inconclusive and inconsistent. The report also documents Kenley's alcohol and drug problems and two potential suicide attempts, one in the military and one drug overdose. The report documents one violent incident during Kenley's incarceration.

Only combined with our summary here would the findings of the state and district courts even begin to portray an accurate picture of the information which counsel reviewed. Counsel testified that he viewed the possible psychiatric and medical testimony as mitigating but felt it was simply too old and inadequate to overcome Dr. Parwatikar's conclusive findings. Counsel apparently missed the references in these materials to the more recent evidence of Dr. Richards's report and Kenley's military record. Each was more recent and more mitigating than Dr. Manion's report. Counsel should have been aware of these as they were clearly documented in one or more places in the information he considered.

The record before us contains Dr. Walker's 1977 report. It states that Kenley has "possible organicity" and suffers from an anti-social personality disorder. The report also documents Kenley's drug abuse and states that there is some indication of pathology possibly contributing to anxiety or schizoid thinking. The report documents an abnormal EEG which "suggests a possible convulsive disorder" and MMPI test results with "peaks on the psychopathic deviate, manic and paranoid scales."

Kenley's military record documents a wrist slitting incident. The military medical records labeled the incident as both a "suicide gesture" and "suicide attempt." The medical evaluation was that Kenley had "suicidal tendencies" and a personality disorder. The record suggests Kenley attempted suicide as a way to get out of the military, a fact Kenley confirmed to Parwatikar. After the wrist slitting incident he was reassigned and was eventually discharged as unsuitable for military service.

Had counsel merely reviewed the information he had and requested the additional reports, he would have found additional documentation of Kenley's problems so as to bolster Manion's findings and further challenge the conclusiveness of Parwatikar's report. As it was, counsel only became aware of Kenley's social history after Crownover initiated contact with him. In fact, the only investigation initiated by counsel was his request for the prison psychiatric examination and his discussion with a non-examining doctor.

In addition to this expert evidence, certain family members had even more recent contact with Kenley and had continued to observe problems and behavior consistent with that documented by Manion, Richards and the military doctors. Family references were contained in Kenley's social history, some were allegedly contacted by Parwatikar, others were discussed with counsel, but none were contacted by counsel. Counsel testified, and the State alleges, that family members were not contacted or called as witnesses because Kenley instructed counsel not to do so. The transcript excerpts in the record only clearly demonstrate that Kenley instructed counsel not to call his father and grandmother.6 Kenley's case is distinguishable from those where a defendant instructs counsel not to call certain witnesses and counsel has no way of knowing about other possible witnesses, Bolder, 921 F.2d at 1363-65, as well as those cases where witnesses were contacted, were not willing to testify favorably, or were unable to provide any material mitigating testimony. Id. (several witnesses contacted); Guinan, 909 F.2d at 1231 (family witnesses were not called but were interviewed); Stokes v. Armontrout, 851 F.2d 1085, 1095-96 (8th Cir.1988) ("not enough to merely speculate that a tearful parent on the stand might evoke pity"), cert. denied, 488 U.S. 1019, 109 S. Ct. 823, 102 L. Ed. 2d 812 (1989).

The affidavits included in the district court file contain statements by roughly a dozen family members further corroborating the facts of Kenley's disadvantaged childhood and his drinking problems, as well as his good qualities such as trustworthiness, work ethic, and so forth. Two of the affiants state they approached counsel with their willingness to provide information about Kenley and sympathetic testimony on Kenley's behalf at trial, but counsel declined. Several affiants also state they saw Kenley the morning after his arrest at the jail and that he was drunk. This fact is corroborated by articles based on witness accounts appearing in the Daily American Republic and a statement made by Sheriff Link of Butler County to counsel before trial that he saw Kenley in a tavern the afternoon before the crimes and that he had second-hand knowledge that Kenley purchased alcohol the night of the crimes.

Turning to counsel's other justification for not presenting available mitigating evidence, that Parwatikar's report was conclusively non-mitigating, if counsel would have reviewed the report closely, he would have found it not so conclusive after all. This is so even without regard to the additional undiscovered reports and witnesses. Parwatikar's report states that he based his findings on the examination he performed, the police report, records of Kenley's 1977 admission to Farmington State Hospital, and interviews with Kenley's family. However, Parwatikar's report contains no reference to Manion or her diagnosis, or to Richards whose 1977 report Parwatikar states he reviewed. The report also indicates Parwatikar knew of Kenley's military record through discussions with Kenley, not from having reviewed the record itself. Parwatikar's report states family was contacted, but no names are listed, no affiants have stated as much, and the record contains no evidence of such contact. The report is inconclusive in two of the three test categories of the examination and Parwatikar's personality diagnosis is not inconsistent with prior diagnoses.

Due to the omissions and incomplete test results, Parwatikar's report can hardly be said to be conclusively non-mitigating. An incomplete or inconclusive report does not obviate the possible significance of older evidence just because it is more recent. Thomas, 738 F.2d at 309 (such a report only found to somewhat reduce the probability of prejudice). The Thomas decision has other similarities to the present case. Thomas's counsel relied solely on his interviews with Thomas and his review of defendant's social history. Thomas had been in the mental hospital on five different occasions, had attempted to overdose on drugs, and was a psychiatric outpatient. This court found ineffective assistance of counsel due to an inadequate investigation. Id. at 308. Kenley's counsel did not do much more than Thomas's counsel, yet Kenley's background was more substantial and better documented.

The fact that Parwatikar's report rules out a mental disease or defect and incompetency does not mean it rules out lesser but potentially mitigating conditions and disorders. We are aware of no legal authority strictly limiting mitigating medical, psychiatric and psychological evidence to that of legal insanity or incompetence. In fact, evidence of conditions, disorders and disturbances are precisely the kinds of facts which may be considered by a jury as mitigating evidence. See Eddings v. Oklahoma, 455 U.S. 104, 114-15, 102 S. Ct. 869, 877, 71 L. Ed. 2d 1 (1982).

None of the information we have discussed was hidden from counsel. References were made to it or it was brought to his attention in the course of his review of materials and his representation of Kenley. We believe that the evidence considered by counsel, along with the evidence that would have been discovered through further investigation, contained much potentially mitigating evidence. Hitchcock v. Dugger, 481 U.S. 393, 398, 107 S. Ct. 1821, 1824, 95 L. Ed. 2d 347 (1987) (evidence consisting of family testimony, troubled and difficult background, prior criminal history or lack thereof, and prospects for rehabilitation are all admissible as mitigating evidence); Eddings, 455 U.S. at 113-16, 102 S.Ct. at 876-78 (facts about family background, personal history, emotional disorders, age, emotional disturbance, beatings by one's father, and slow mental development are admissible mitigating evidence). Thus, this case may be distinguished from one like Laws where the court found that defendant's records revealed no mitigating medical or psychiatric problems.

When counsel makes a decision not to further investigate, that decision is only reasonable to the extent professional judgment makes the limitations placed on further investigation reasonable in the circumstances. Strickland, 466 U.S. at 690-91, 104 S. Ct. at 2066. We will not fault a reasonable strategy not to investigate further if it is based on sound assumptions. Pickens, 714 F.2d at 1467. Here it was not a reasonable strategy that led counsel not to investigate, but lack of thoroughness and preparation. "Counsel can hardly be said to have made a strategic choice against pursuing a certain line of investigation when s/he has not yet obtained the facts on which such a decision could be made." Chambers, 907 F.2d at 835 (quoting United States v. Gray, 878 F.2d 702, 711 (3rd Cir.1989)).

In sum, the state court fact findings were incomplete and counsel's decision to stop investigating was unreasonable because his belief that mitigating evidence was too old and insubstantial resulted from his failure to follow available leads to more recent and persuasive mitigating evidence. Counsel's investigation fell short of what was required in this case. We also feel counsel unreasonably concluded Parwatikar's report was conclusively non-mitigating because the report itself was incomplete, often inconclusive, and did not rule out the possibility of mitigating evidence of disorders, conditions and disturbances.

Having found counsel's performance deficient for failing to further investigate available and potentially mitigating evidence, we must now decide whether counsel's performance prejudiced Kenley. The standard by which we judge prejudice is whether we find a reasonable probability that the outcome of the sentencing proceeding would have been different but for counsel's ineffectiveness. A reasonable probability is one sufficient to undermine our confidence in the outcome of the proceeding. We believe there is a reasonable probability Kenley would not have received a death sentence if some of the mitigating evidence had been presented.

The state courts and district court concluded as a matter of law that there was no prejudice because the mitigating evidence was "scant" and the aggravating evidence would have been more damaging. We have demonstrated that the state court fact finding was not supported by the record because the potential mitigating evidence was substantial. We do not agree as a matter of law that there was no prejudice because the other evidence would have significantly worsened the perception of Kenley's character. Rather, we believe the testimony would have put the aggravating evidence in context along with the mitigating evidence.

The state courts' conclusion on this issue was based on counsel's beliefs that the mitigating evidence would be unpersuasive to a jury in light of Dr. Parwatikar's report and additional potentially aggravating information. We do not believe reasonably effective counsel would have reached the same conclusion if, in addition to reviewing Crownover's records, counsel would have interviewed Dr. Manion and Dr. Richards or at least reviewed their reports, reviewed Kenley's military file, and interviewed family witnesses concerning Kenley's history and his condition the morning after arrest. This additional information would have been more than enough to mount a challenge to Parwatikar's report. The medical evidence of Kenley's condition may fall short of a mental disease or defect diagnosis but it is more than just the borderline non-severe personality disorder we found in Smith v. Armontrout, 888 F.2d 530, 534 (8th Cir.1989), and is more akin to an extreme personality or emotional disorder or disturbance. Cf. Strickland, 466 U.S. at 700, 104 S.Ct. at 2071 (at most evidence showed defendant was generally a good person, under a lot of stress, but not suffering from an extreme disturbance of any kind). Also, Kenley's alcohol problems were well documented by all doctors, and when combined with other evidence from the sheriff, news accounts and family testimony, would have been probative mitigating evidence of intoxication.

With regard to the potential for additional aggravating information, the most damning information in Parwatikar's report is the determination that Kenley was not insane or incompetent. As stated, as far as we are aware, such conditions are not prerequisites to information being considered mitigating evidence. See Eddings, 455 U.S. at 114-15, 102 S.Ct. at 877; Hayes v. Lockhart, 852 F.2d 339, 349 (8th Cir.1988) (psychiatric report stating defendant did not have mental disease or defect but documenting alcohol problem and anti-social personality disorder admitted as mitigating evidence), rev'd on other grounds, 491 U.S. 902, 109 S. Ct. 3181, 105 L. Ed. 2d 691 (1989).

The other potentially damaging information consists principally of Kenley's anti-social behavior, violence and arrests (not convictions) when he was a minor. His sole conviction was already in evidence. Had all the reports been introduced into evidence and had all affiants and doctors testified, we do not deny the jury would have heard aggravating information. After conducting a reasonable investigation, counsel may conclude that the introduction of the mitigating evidence "would barely have altered the sentencing profile presented", Strickland, 466 U.S. at 700, 104 S.Ct. at 2071, or that harmful information is greater than potential helpful information, and choose not to present certain evidence. Burger v. Kemp, 483 U.S. 776, 789-91, 107 S. Ct. 3114, 3123-24, 97 L. Ed. 2d 638 (1987); Guinan, 909 F.2d at 1231. In this case we have concluded counsel did not conduct a reasonable investigation. But even if he had, in the context of what the jury had already heard at trial about Kenley's crazed and violent actions on the night of the crimes, we are doubtful that much of the additional aggravating information would have had any significant incremental aggravating effect on the jury. Furthermore, the aggravating information was mostly cumulative. Chambers, 907 F.2d at 829-30.

Given the sympathetic light in which Kenley's past behavior could have been presented, in the context of his family and medical background by Crownover and family affiants, given the significant evidence of medically significant conditions and disorders, suicidal tendencies, alcohol abuse and intoxication, and the prospects for successful treatment, we believe it was unreasonable for counsel to have presented no mitigating evidence.

Reasonable counsel might have been somewhat selective in his evidence presentation, but would not have cast aside all the mitigating evidence in this case. See Guinan, 909 F.2d at 1229 (after thoroughly reviewing all medical and psychiatric records, counsel requested a second psychiatric opinion); Smith v. Armontrout, 888 F.2d at 535 (decision by counsel to offer a mitigating circumstance, "support it with expert testimony, but leave out documents that would at best have brought some additional increment of support for the diagnosis, while also carrying a substantial risk" of making things worse, not unreasonable); Hayes 852 F.2d at 348-51 (counsel not ineffective for introducing only part of medical and psychological evidence where counsel conducted a thorough investigation, did introduce evidence of alcohol problems and treatment prospects, called live witnesses, and introduced psychiatric report into evidence). For the reasons stated, we find prejudice.

The State argues and the district court infers that recent case authority in our circuit does not support the legal conclusions we reach in this case, and suggests our existing authority creates an almost insurmountable barrier to habeas relief. We disagree with this general characterization of our case authorities although, as a practical matter, the strong presumption in favor of counsel and the heavy burden on the defendant established by the Supreme Court in Strickland means determinations of ineffective assistance of counsel such as in the present case will likely occur infrequently.

Having found deficient performance and prejudice in the sentencing phase of the present case, we conclude counsel was ineffective. We therefore reverse so much of the judgment of the district court as denied the writ with respect to the death sentence and remand this case to the district court with instructions that the State of Missouri be required to vacate Kenley's death sentence and either reduce it to a life sentence without parole or conduct a new sentencing procedure within one hundred eighty days from and after the issuance of the district court's order. See Mo.Stat.Ann. Sec. 565.035, subd. 5(2), (3) (1983).