Appellant Jack Dixon appeals the district court's affirmance of his termination of disability benefits. We reverse and remand for further development of the factual record.

Appellant Jack Dixon was born on August 21, 1948. He received a ninth grade education and worked as a truck driver and heavy equipment operator. Dixon suffered a job related injury on January 9, 1989, when he fell off a loading dock and injured his tail bone, resulting in an L1 vertebral body compression fracture, paralumbar muscle strain, myositis with myospasticity, and a coccygeal contusion. Dr. Hermie Plunk, a primary care physician, was Dixon's treating physician. Dr. Plunk has continued to treat Dixon throughout this appeal. Dixon also had trouble with his vision for a number of years. Dixon received treatment for his vision problems from Dr. Roger Baker, an ophthalmologist. Dr. Baker concluded Dixon had only hand motion vision in the right eye, and 20/25 vision in the left eye.

In June of 1989, Dixon filed an application for disability insurance benefits and for supplemental security income benefits alleging a disability onset date of January 9, 1989, due to an injured tail bone and poor vision. On May 3, 1990, Dixon was awarded benefits on his claim with an onset date of January 9, 1989.

In 1994, while Dixon was still receiving disability benefits based on his back and vision problems, he began to have heart problems. Dr. Fraser Richards successfully performed coronary artery bypass surgery on Dixon on June 1, 1994. Dixon continued to see his primary care physician, Dr. Plunk, for cardiac problems, hypertension, and anxiety, while also seeing Dr. Richards for his cardiac problems. From June 21, 1994, through December 17, 1996, Dr. Plunk saw Dixon a number of times for cardiac and non-cardiac symptoms and for medication management.

Dixon was admitted to the hospital on May 24, 1997, for chest pain. While in the hospital, a cardiac catheterization revealed coronary atherosclerotic disease ranging from mild stenosis in some arteries to severe stenosis or total occlusion in others. Tests performed on Dixon revealed he had sinus bradycardia. He was diagnosed with non-cardiac chest pain, coronary artery disease, hypertension, and a history of tobacco abuse. Dixon was discharged on May 26, 1997, with a medication regime that included Ecotrin, Nitroglycerin, Norvasc, and Xanax. The medication regime was altered throughout 1997 by Dr. Plunk and Dr. Richards as Dixon continued to experience symptoms. A June 13, 1997, stress test revealed that Dixon failed to achieve the target heart rate and that he had abnormal blood pressure.

The Social Security Administration continued to review Dixon's disability award through the continuing disability review process. A March 17, 1998, report by Dr. Plunk, performed at the request of the Social Security Administration, stated that Dixon's physical ability to perform activities of daily living were limited by coronary artery disease, post CASG triple bypass surgery, unstable angina, hypertension, hystoplasmosis, hiatal hernia, difficulty hearing, and decreasing vision. Dr. Richards, Dixon's treating cardiac surgeon, gave no opinion on Dixon's work-related or daily living activities.

In a letter dated April 26, 2000, Dr. Plunk reiterated her previous diagnosis and concluded: "[Dixon's] overall health condition at this time is weakened yet stable with medication. He cannot work in extreme heat or cold. He cannot lift over ten pounds. He cannot return to his previous occupation in construction as a heavy equipment operator nor truck driver. I do not expect his condition to improve."

On May 3, 1998, Dr. Robert Redd reviewed the medical evidence in Dixon's record to make a capacity assessment2. Dr. Redd did not treat nor did he examine Dixon. Dr. Redd did not have the statements of Dr. Plunk or Dr. Swingle regarding Dixon's work limitations. Dr. Redd concluded that Dixon had a medium level residual functional capacity (RFC) reduced by vision-related limitations.

Dixon also underwent, at the request of the Social Security Administration, a consultative examination performed by Dr. Charles Swingle on May 5, 1998. Dr. Swingle diagnosed Dixon with post-by-pass coronary artery disease, hypertension, blindness in the right eye, hiatal hernia, and histoplasmosis. Dr. Swingle stated that Dixon was "probably unable to be gainfully employed due to the severity of [coronary artery disease] and blindness in [the] right eye."

Dr. F. Joseph George, with the Jonesboro Eye Clinic, also treated Dixon in 1998 and 1999. Dr. George concluded that Dixon's best corrected central visual acuity was the ability to count fingers with the right eye and 20/20 vision in the left eye. This represented a slight improvement in Dixon's left eye vision. Dr. George did not believe that the poor vision in the right eye could be improved with treatment. Dr. George did note early cataract changes in each eye.

Pursuant to the continuing disability review process, Dixon received a notice of disability cessation on May 18, 1998. The communication notified Dixon that he was no longer disabled as of May 1, 1998, and that his final benefits payment would be made on July 31, 1998. This decision was affirmed on reconsideration before a disability hearing officer on June 24, 1999. On June 27, 2000, an ALJ upheld the determination that Dixon was no longer disabled within the meaning of the Social Security Act. Dixon was not represented by counsel at the hearing. The Appeals Council declined to review the decision. After retaining counsel, Dixon sought judicial review of the ALJ's final decision and on March 28, 2002, a United States Magistrate Judge affirmed the ALJ's determination that Dixon was no longer disabled. On appeal, Dixon contends the ALJ's decision was not supported by substantial evidence.

We review the denial of benefits pursuant to the continuing disability review process for substantial record evidence to support the ALJ's decision. See Muncy v. Apfel, 247 F.3d 728, 730 (8th Cir.2001). Substantial evidence is relevant evidence that reasonable minds might accept as adequate to support the decision. Hunt v. Massanari, 250 F.3d 622, 623 (8th Cir.2001). Therefore, the narrow issue on appeal is whether there is substantial record evidence to support the ALJ's determination that Dixon could perform medium work.

The continuing disability review process is a sequential analysis prescribed in 20 C.F.R. § 404.1594(f). The regulations for determining whether a claimant's disability has ceased may involve up to eight steps in which the Commissioner must determine (1) whether the claimant is currently engaging in substantial gainful activity, (2) if not, whether the disability continues because the claimant's impairments meet or equal the severity of a listed impairment, (3) whether there has been a medical improvement, (4) if there has been medical improvement, whether it is related to the claimant's ability to work, (5) if there has been no medical improvement or if the medical improvement is not related to the claimant's ability to work, whether any exception to medical improvement applies, (6) if there is medical improvement and it is shown to be related to the claimant's ability to work, whether all of the claimant's current impairments in combination are severe, (7) if the current impairment or combination of impairments is severe, whether the claimant has the residual functional capacity to perform any of his past relevant work activity, and (8) if the claimant is unable to do work performed in the past, whether the claimant can perform other work. 20 C.F.R. § 404.1594(f).

The ALJ first determined that Dixon had not been engaged in substantial gainful work activity. In the next step, the ALJ concluded that "[w]hile the claimant's current conditions and/or impairments ... are potentially severe within the meaning of the Regulations, there is no evidence in this claim to show that they have been so pervasive and/or severe as to meet or medically equal one of the impairments listed in Appendix 1, Subpart P, Regulations No. 4." ALJ Decision, July 27, 2000, at 3. The ALJ continued in the sequential analysis and determined that there was medical improvement in Dixon's impairments. Next, the ALJ found that Dixon's medical improvement was related to his ability to do work and his RFC had increased to at least the medium duty level3. The ALJ then determined that Dixon's current impairments were still severe, but that he had the RFC to perform medium work. Specifically, the ALJ concluded:

In assessing the claimant's residual functional capacity consideration is given to the above factors regarding the wide range of daily activities performed by him including driving, gardening, mowing, shopping, and reading which discredit the assessment reached by Dr. Plunk in his April 26, 2000 correspondence. Testimony disclosed he rarely experiences back discomfort and treatment notes disclose a negative cardiolite study in July 1999. Based upon these considerations he is assessed residual functional capacity to lift/carry 50 pounds occasionally and 25 pounds frequently; stand/walk 6 hours in an 8 hour day; sit 6 hours in an 8 hour day; and push/pull using the upper and lower extremities.

ALJ Decision, July 27, 2000, at 5. Finally, the ALJ determined that Dixon could not perform his past relevant work, but, according to testimony from a vocational expert, there were jobs within the economy which Dixon could perform.

The ALJ based his decision on three principal factors. First, the ALJ correctly noted that Dixon's back pain had improved and that Dixon rarely experienced back discomfort. Second, the ALJ found persuasive the fact that Dixon was able to engage in a wide range of certain daily activities, such as driving, gardening, mowing, shopping, and reading. The ALJ believed these activities discredited the assessment reached by Dixon's treating physician, Dr. Plunk, and his examining physician, Dr. Swingle. Third, the ALJ relied on the negative cardiolite study in 1999 to support his determination that Dixon could perform medium work. We address these factors in turn.

The ALJ determined that Dixon rarely experienced back discomfort. It is undisputed that Dixon's back problems have improved and we agree there is substantial record evidence to support that determination. However, the mere fact that Dixon's back problems have improved does not mean, in and of itself, that Dixon can do medium work. Dixon still suffers from poor vision and the record indicates a significant history of cardiac problems since the initial onset of his disability and during his award period. While the record was well-developed regarding Dixon's back pain, we find the record insufficiently developed regarding Dixon's cardiac problem and the impact it has on his ability to work.

In finding Dixon capable of performing the exertional tasks of medium work, the ALJ relied on the statements in the record that Dixon engaged in daily life activities of driving, gardening, mowing, shopping, and reading. The ALJ believed these activities discredited the assessment reached by Dr. Plunk, Dixon's treating physician, that Dixon had a lifting limitation of ten pounds. However, the record pertaining to Dixon's daily activities was not fully developed. There is nothing in the record to support the conclusion that because Dixon can perform certain daily life activities, he can also perform the exertional tasks of medium work. "This court often has noted ... that a claimant's ability to perform household chores does not necessarily prove that claimant capable of full-time employment." See Ekeland v. Bowen, 899 F.2d 719, 722 (8th Cir.1990) (citing Easter v. Bowen, 867 F.2d 1128, 1130 (8th Cir.1989)). The fact that Dixon usually walks two miles a day at a slow pace does not, in itself, mean Dixon can perform medium work. The ALJ failed to explain how Dixon's daily life activities translate to an ability to perform the exertional tasks required in medium work.

The Commissioner contends that the ALJ adopted most of Dr. Plunk's assessment, but did not agree with the ten pound limitation because the results of his cardiolite test do not support that conclusion. The ALJ did not explain the significance of the negative cardiolite test, nor is its significance developed in the record. The test is mentioned in the cardiologist's report without any explanation of its implications. The record needs to be more fully developed regarding what specifically the cardiolite test results, among other information, means relative to Dixon's ability to work, and how those particular test results may conflict with other tests in Dixon's medical records. Absent that information, it is not possible to ascertain Dixon's ability to work without engaging in medical conjecture. See Nevland v. Apfel, 204 F.3d 853, 858 (8th Cir.2000) ("`An administrative law judge may not draw upon his own inferences from medical reports.'") (quoting Lund v. Weinberger, 520 F.2d 782, 785 (8th Cir.1975)). On the record as it stands, the results of the cardiolite test provide no basis for rejecting the recommendations of Dr. Plunk and Dr. Swingle.

Finally, we find merit in Dixon's contention that the ALJ relied too heavily on the opinion of the Social Security Administration's reviewing physician, Dr. Redd. While the ALJ's decision does not mention Dr. Redd's recommendation, it is record evidence. Dixon contends Dr. Redd's report is the only evidence in the record which supports the ALJ's RFC determination that Dixon can perform medium work. Dr. Redd never examined Dixon and provided his RFC opinion based solely on Dixon's medical records. See Nevland, 204 F.3d at 858 (relying upon non-examining, non-treating physicians to form an opinion on a claimant's RFC does not satisfy the ALJ's duty to fully and fairly develop the record). The record indicates that Dr. Redd did not have Dr. Plunk's or Dr. Swingle's recommendations and statements to inform his RFC recommendation. We have previously stated that "[a] treating physician's opinion should be accorded substantial weight." Prince v. Bowen, 894 F.2d 283, 285 (8th Cir.1990) (citations omitted). Dixon's treating physician and the state's examining physician concluded that Dixon had, at a minimum, more severe work restrictions than what the ALJ determined. Dr. Redd concluded that Dixon could perform medium work. Given the contradicting recommendations in the record and the insufficiently developed record surrounding Dixon's cardiac problems, Dr. Redd's opinion does not constitute substantial record evidence that Dixon can perform medium work. See Nevland, 204 F.3d at 858 ("The opinions of doctors who have not examined the claimant ordinarily do not constitute substantial evidence on the record as a whole.").

The ALJ's conclusions were not supported by substantial record evidence. Specifically, the record is undeveloped regarding: Dixon's daily life activities and how those activities relate to his ability to work; the significance of certain test results relied upon by the ALJ; and, what specifically Dixon's treating and examining physicians recommend in terms of his RFC. A more complete record needs to be developed to ascertain what level of work, if any, Dixon is able to perform. Accordingly, we remand the case for further proceedings to more fully develop the record.