Claimant Nick L. Blea appeals from a magistrate judge’s order affirming the Commissioner’s denial of his application for disability insurance benefits under Title II of the Social Security Act. Mr. Blea contends that the magistrate judge erred in upholding the administrative, law judge (ALJ)’s decision, where the ALJ failed to: (1) utilize Social Security Ruling (SSR) 83-20 to determine the onset date of Mr. Blea’s disabilities, and (2) confront or address lay witness testimony pursuant to Social Security Rulings 83-20 and 85-16. We hold the onset date of Mr. Blea’s disabilities was ambiguous, that the ALJ erred by failing to utilize SSR 83-20 to infer an onset date, and that the ALJ should have addressed lay witness testimony in his written decision. Therefore, exercising jurisdiction pursuant to 42 U.S.C. § 405(g) and 28 U.S.C. § 1291, we reverse and remand for further proceedings consistent with this opinion.
I. BACKGROUND
In July 1997, Mr. Blea’s brother shot him in the foot. Mr. Blea endured a series of seven surgeries over the next two months, and numerous visits with doctors thereafter. In January 1998, Mr. Blea’s treating physician, Dr. William L. Ritchie, M.D., provided him with a handicapped parking placard, and wrote in his treatment notes that Mr. Blea’s “pain is probable [sic] due to the bony deformities, although there is the possibility that it is due to incomplete fusion of the bone graft.” Aplt’s App. vol. I, at 161. Dr. Ritchie also noted that despite the pain, Mr. Blea was “able to ambulate -with minimal restrictions.” Id.
In February, Dr. Ritchie again saw Mr. Blea, who “return[ed] complaining of continued foot pain.” Id. at 159. Dr. Ritchie prescribed ibuprofen and Darvocet-N, advised Mr. Blea to use a cane, and referred him to Dr. Laura A. Mitchell, M.D., a foot and ankle specialist. After examining him in March, Dr. Mitchell noted that his x-rays revealed a “lateral shift of the lesser metatarsals and some shortening of the first ray.” Id. at 157. She further explained that Mr. Blea was “doing well and is not interested in further surgery.” Id. at 156. Mr. Blea did not see another physician until June 1998, when he tripped while carrying his child. That doctor diagnosed Mr. Blea with a broken toe, and noted that “[h]e has no pain proximally where he had the previous bone graft.” Id. at 154. Although the doctor he saw at this time recommended that Mr. Blea make an appointment to follow up with Dr. Ritchie in a few weeks, Mr. Blea did not seek medical treatment again until January 2000.
At the January 2000 visit, Dr. Ritchie noted that the wounds from surgery and the injury itself were “well healed,” Id. at 153. The x-rays showed “some degenera*906tive changes” in Mr. Blea’s foot; however, “no acute changes [were] noted.” Id. He recommended that Mr. Blea see Dr. Mitchell again. But, at his February 2000 visit, Dr. Mitchell diagnosed Mr. Blea with “significant post-traumatic arthritis with constant pain.” Id. at 150. She discussed the possibility of a Syme’s amputation — an amputation of the entire foot at the ankle — with Mr. Blea at this visit. Further, in February 2002, Mr. Blea saw Dr. Robert C. Schenck, M.D., who indicated that Mr. Blea’s x-rays “reveal[ed] fairly significant midfoot and forefoot arthritis.” Id. at 197. Dr. Schenck also recommended that Mr. Blea apply for social security benefits: “in my opinion, he is permanently fully disabled and unemployable.” Id. at 198.
Mr. Blea followed Dr. Schenck’s advice. In March 2002, he applied for disability insurance benefits under Title II of the Social Security Act and supplemental social security income under Title XVI, alleging he was disabled due to post-traumatic arthritis and depression as of June 1997. While his claim for benefits was pending, Mr. Blea met with Dr. Elegió R. Padilla, Ph.D., on August 28, 2002, one of the Social Security Administration’s consultative psychologists. This was the first time Mr. Blea received any medical treatment for his depression. During the evaluation, Mr. Blea revealed his past suicidal thoughts, described his problems with alcohol, and discussed the emotional ramifications of having been shot by his own brother. Dr. Padilla concluded that Mr. Blea “has been dysthymic for years and may be suffering from a major depression currently.” Id. at 246.
The Commissioner initially denied Mr. Blea’s claims for disability insurance and supplemental income. Upon reconsideration in October 2002, however, Mr. Blea was found to be disabled and entitled to supplemental security income as of March 1, 2002. “The decision was made on the basis that at the time the application was protectively filed, March 9, 2002, the claimant’s post traumatic arthritis and dysthymia were medically determinable impairments that were disabling.” Id. at 18. Thus, the Commissioner determined that, as of at least March 1, 2002, Mr. Blea was disabled due to both post-traumatic arthritis and dysthymia. Nonetheless, the Commissioner denied Mr. Blea’s application for disability insurance because she determined that any impairments Mr. Blea had were “not disabling on any date through 12/31/98 the last day insured status for disability was met.” Id. at 42.
After the denial of his claim for disability insurance benefits, Mr. Blea requested a hearing before an ALJ, which was held in September 2003. During the hearing, Mr. Blea testified about the effects of having been shot in the foot. Mrs. Blea also testified. She reported that after the gunshot injury and for months thereafter, her husband talked about suicide. Even recently, she testified, “he talks about it probably at least two to three times a week ... he says that the pain is too much to bear.” Id. at 333. She also explained that since the gunshot injury, Mr. Blea has rarely left the house due to pain and anxiety, and has also been unable to sleep comfortably. At the close of the evidence, Mr. Blea’s attorney asked that the ALJ call a medical expert to assist the ALJ in reviewing the medical evidence. The ALJ denied this request because he found “no medical — or no psychiatric evidence relating to [the time before Dec. 31, 1998] which [he] feel[s] warrants that review.” Id. at 342.
In November 2003, the ALJ denied Mr. Blea’s claim for disability insurance benefits because he decided that Mr. Blea had not been disabled as of the date he last was eligible for disability benefits, Decern*907ber 31, 1998. The ALJ did not, however, determine an exact onset date. Rather, the ALJ applied the five-part sequential analysis mandated by 20 C.F.R. §§ 404.1520 and 416.920, and determined that because Mr. Blea retained the capacity for sedentary work as of December 31, 1998, he could not have been disabled.
With respect to Mr. Blea’s depression, the ALJ found that Mr. Blea’s allegations about his physical limitations due to the condition were not well supported by the medical record. Additionally, his reports of depression were not credible before December 31, 1998 since there was no evidence of treatment or reports of depression before that date. Thus, the ALJ did not consider Mr. Blea’s claim of depression in determining what kind of residual capacity for work Mr. Blea retained.
With respect to his post-traumatic arthritis, the ALJ applied the five-step sequential analysis pursuant to 20 C.F.R. §§ 404.1520 and 416.920 to determine whether Mr. Blea was disabled as of that date. Mr. Blea carried the burden of proof with respect to the first four parts of the test.
The first part inquires whether the claimant has performed substantial gainful activity since the alleged onset date; if he has, he is not disabled. The ALJ found that Mr. Blea had not performed substantial gainful activity since the gunshot wound. The second part asks whether the claimant’s impairments are severe, which they must be if he is to be found disabled. The ALJ found that Mr. Blea’s impairment due to post-traumatic arthritis was severe.
If the claimant has not performed gainful activity and his impairments are severe, then the ALJ must determine whether his impairments are listed in the Regulations. If the impairments are listed, then the ALJ must find the claimant disabled without further inquiry. At this step, the ALJ determined that “[t]he claimant’s musculoskeletal impairments did not approach the severity required to meet listings, § 1.00 prior to December 1998. Mr. Blea retained the ability to ambulate effectively....” Aplt’s App. at 21. Thus, the ALJ proceeded to the fourth step of the inquiry, whether the impairments prevent Mr. Blea from performing his past relevant work. The ALJ found that after the gunshot injury, Mr. Blea could no longer perform his relevant past work, even before December 31, 1998.
Once the claimant has carried his burden through the first four parts, the burden shifts to the Commissioner for the fifth and final part, to show that other work, which the claimant could perform given his residual functional capacity, existed in significant numbers in the national economy such that the claimant could have been employed. If the Commissioner can make this showing, then the ALJ must find that the claimant is not disabled. The ALJ found that as of December 31, 1998 Mr. Blea retained the ability to do sedentary work, relying heavily on the testimony of an impartial vocational expert. The ALJ additionally noted that, in March 1998, Mr. Blea and his doctors were satisfied with the results of his surgeries, in June 1998, he walked well enough to carry a baby, and he did not seek medical treatment again until January 2000. Based on these factors, the ALJ decided that it was reasonable to believe that Mr. Blea did not have severe, debilitating pain during the relevant time period. Therefore, the ALJ concluded that Mr. Blea’s allegations of complete disability due to post-traumatic arthritis were not supported by the medical record before his last insured date of December 31,1998.
Mr. Blea requested the Appeals Council to reverse the ALJ, providing additional *908evidence from one of his treating orthopedists, Dr. Schenck. Dr. Schenck offered the opinion that “it is reasonable to assume that Mr. Blea’s post-traumatic arthritis with constant pain was present between June 1998 and January 2000.” See Aplt’s App. at 309. The Appeals Council, however, declined review, stating in a form decision that it had “considered the additional evidence listed.” Id. at 6. When the Appeals Council declined review, the ALJ’s decision became the final decision of the Commissioner.
Mr. Blea appealed the ALJ’s decision to the federal district court, arguing that the ALJ erred by not applying SSR 83-20 to determine the onset date of his disabilities. He also contended that the ALJ’s failure to address the evidence provided by lay witnesses constituted reversible error. A magistrate judge issued judgment pursuant to 28 U.S.C. § 636(c), holding that the ALJ’s decision was supported by substantial evidence. This appeal followed.
II. DISCUSSION
‘We review the [magistrate judge’s] decision de novo and independently determine whether the ALJ’s decision is free from legal error and supported by substantial evidence.” Fischer-Ross v. Barnhart, 431 F.3d 729, 731 (10th Cir.2005). Substantial evidence is “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Fowler v. Bowen, 876 F.2d 1451, 1453 (10th Cir.1989) (internal quotation marks omitted). We will not reweigh the evidence or substitute our judgment for the Commissioner’s. Qualls v. Apfel, 206 F.3d 1368, 1371 (10th Cir.2000). “Because the Appeals Council denied review, the ALJ’s decision is the Commissioner’s final decision for purposes of this appeal.” Madrid v. Barnhart, 447 F.3d 788, 789-90 (10th Cir.2006). We note that any new evidence submitted to the Appeals Council on review “becomes part of the administrative record to be considered when evaluating the Secretary’s decision for substantial evidence.” O’Dell v. Shalala, 44 F.3d 855, 859 (10th Cir.1994) (reasoning that “because the Secretary’s decision does not become final until after the Appeals Council denies review or issues its own findings, her ‘final decision’ necessarily includes the Appeals Council’s conclusion that the ALJ’s findings remained correct despite the new evidence.”).
Mr. Blea presents two main arguments on appeal. First, he contends that the ALJ should have utilized Social Security Ruling 83-20, Program Policy Statement: Title II and XVI: Onset of Disability (PPS-100), 1983 WL 31249 (S.S.A. 1983)(“SSR 83-20”) to determine the onset date of his disabilities. Second, Mr. Blea argues that the ALJ’s failure to analyze Mrs. Blea’s lay witness testimony constitutes reversible error, citing Tenth Circuit precedent, as well as the regulations SSR 83-20, and Social Security Ruling 85-16, Program Policy Statement: Titles II and XVI: Residual Functional Capacity for Mental Impairments (PPS-120), 1985 WL 56855 (S.S.A.1985)(“SSR 85-16”), which instruct an ALJ to consider lay witness testimony. We examine each argument below.
A. Application of SSR 83-20
To qualify for disability benefits, a claimant must establish that he is “disabled” under the Social Security Act, 42 U.S.C. § 423(a)(1)(E) (the “Act”). The Act states that “disability” is the “inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment which ... has lasted or can be expected to last for a continuous period of not less than 12 months.” Id. § 423(d)(1)(A). It is not *909disputed that Mr. Blea is currently disabled under the Act. What is in dispute is when Mr. Blea became disabled. This is problematic for Mr. Blea because he only qualifies for disability benefits if he can show that he was disabled prior to his last insured date — December 31, 1998. See id. § 423(c).
On appeal, Mr. Blea presents two arguments regarding why the ALJ should have applied SSR 83-20 to determine the onset date of his disabilities. First, Mr. Blea contends that his post-traumatic arthritis is of traumatic origin under SSR 83-20 because it was caused by the gunshot wound and presented within twelve months of the gunshot incident, thereby making the onset date of his disability June 1997 — the date of the gunshot injury. Alternatively, he argues that even if his post-traumatic arthritis and dysthymia are not of traumatic origin, the ALJ committed reversible error by failing to follow SSR 83-20’s provisions for calling a medical advisor. Before addressing Mr. Blea’s arguments, we examine SSR 83-20 and discuss what it requires of an ALJ.
1. SSR 83-20
SSR 83-20 is “binding on all components of the Social Security Administration,” including ALJs, 20 C.F.R. § 402.35(b)(1), and sets forth an analytical framework for assessing the date of onset for a disability of traumatic or non-traumatic origin. It provides that a disability is of “traumatic origin,” where after the date of injury, “the individual is thereafter expected to die as a result or expected to be unable to engage in substantial gainful activity (SGA) (or gainful activity) for a continuous period of at least 12 months.” SSR 83-20, at 2. Where a disability is of traumatic origin, the date of onset is the date of the traumatic injury. Id.
Additionally, SSR 83-20 provides a framework for examining injuries that are not considered of “traumatic origin” under the regulation. SSR 83-20 states that “[i]n disabilities of nontraumatic origin, the determination of onset involves consideration of the applicant’s allegations, work history, if any, and the medical and other evidence concerning impairment severity.” Id. The date alleged by the claimant is the starting point for determining disability onset, and the date the claimant stopped working is also of significance in selecting the onset date. Id. Medical evidence, however, is the “primary element” for the onset determination, as the onset date “can never be inconsistent with the medical evidence of record.” Id. at 2-3.
SSR 83-20 also provides that, when medical evidence does not establish the precise onset date, the ALJ may have to “infer the onset date from the medical and other evidence that describe the history and symptomatology of the disease process.” Id. at 2. The regulation provides two examples of situations where it may be necessary to infer an onset date: (1) in the case of a slowly progressing impairment, “when, for example, the alleged onset and the date last worked are far in the past and adequate medical records are not available,” and (2) when “onset of a disabling impairment(s) occurred some time pri- or to the date of the first recorded medical examination.”1 Id. at 3. “At the hearing, *910the [ALJ] should call on the services of a medical advisor when onset must be inferred.” Id.
2. Traumatic Origin Provisions of SSR 83-20
Mr. Blea contends that his disabilities are of traumatic origin under SSR 83-20, and therefore the onset date should be the date he was shot in the foot — June 1997. The ALJ, without explicit consideration of SSR 83-20, did not use this date, finding that “Mr. Blea’s [gunshot] wound initially heeled [sic] well and he displayed minimal pain in the foot.” Aplt’s App. vol. I at 21.
On appeal, Mr. Blea renews his argument that the ALJ’s implicit decision that his disability was not of traumatic origin was unsupported by the medical evidence and contrary to SSR 83-20. He states that “[t]he twelve month requirement under SSR 83-20 was met when, according to the evidence, claimant’s disabling post-traumatic arthritis had developed as a result of his injury of June 1997 and had presented itself by the following year in June 1998.” Aplt’s Br. at 10. The government responds that the medical evidence available from June 1997 to June 1998 shows that Mr. Blea had been steadily improving and could ambulate effectively; thus, the ALJ correctly decided that the onset date of Mr. Blea’s impairments did not correspond to the gunshot injury.
We begin our analysis of Mr. Blea’s argument by noting that SSR 83-20 and our precedent do not require an ALJ to consult with a medical advisor to determine whether a disability is of traumatic origin. We have previously held that “a medical advisor need be called only if the medical evidence of onset is ambiguous.” Reid v. Chater, 71 F.3d 372, 374 (10th Cir.1995). If a disability is of traumatic origin, the onset date would not be ambiguous; indeed, it would most often be “self-evident.” Mason v. Apfel, 2 F.Supp.2d 142, 149 (D.Mass.1998). Thus, our task at this point is to determine if the ALJ’s decision that Mr. Blea’s onset date was not the date of the gunshot injury comported with the traumatic-injury provisions of SSR 83-20 and was supported by substantial evidence.
Mr. Blea asserts that symptoms of his post-traumatic arthritis presented sometime in the twelve months after the gunshot injury, but he does not cite any part of the record that supports this assertion, even though the record contains ample medical evidence for the time period June 1997 to June 1998. Our review of the record shows that at the time of the gunshot wound, and during the healing process thereafter, there were no indications that Mr. Blea was “expected to be unable to engage in substantial gainful activity ... for a continuous period of at least 12 months” due to post-traumatic arthritis or dysthymia. SSR 83-20, at 2.
Although Mr. Blea still suffered from pain, by January 1998, seven months after the injury, he was “able to ambulate with minimal restrictions.” Aplt’s App. vol. I at 161. In March 1998, he sought the advice of a foot and ankle specialist, who prescribed a new orthotic to assist Mr. Blea in walking. The specialist’s notes state that, besides some numbness, Mr. Blea was “doing well and is not interested in any further surgery.” Id. at 156. Finally, a doctor’s notes from June 1998 indicate that Mr. Blea had “no pain proximally where he had the previous bone graft,” supporting the ALJ’s conclusion that Mr. Blea was not exhibiting symptoms of post-traumatic arthritis at that visit. *911Id. at 154. Therefore, we conclude that the ALJ’s decision not to use the date of the gunshot as the onset date did not conflict with SSR 88-20, and that it was supported by substantial evidence.
3. Provisions of SSB 88-20 Relating to Sloiuly Progressing Impairments
The issue of whether Mr. Blea’s disabilities are slowly progressive impairments, where “[d]etermining the proper onset date is particularly difficult,” SSR 83-20 at 3, is a much closer question. Mr. Blea contends that even if the provisions of SSR 83-20 relating to traumatic origin do not apply to this case, the ALJ erred by failing to call a medical advisor to assist in setting an onset date. The Commissioner replies that the ALJ was not obligated to consult with a medical advisor because there was substantial evidence to conclude, at step five of the five-step sequential analysis, that Mr. Blea retained the capacity to perform sedentary work.
We begin by rejecting the Commissioner’s position because it fails to address the crux of the issue. The ALJ’s finding of residual functional capacity at step five does not mean that the ALJ can ignore the clear directives of SSR 83-20, which is “binding on all components of the Social Security Administration.” 20 C.F.R. § 402.35(b)(1).
It is important to understand that the issue of whether a medical advisor is required under SSR 83-20 does not turn on whether the ALJ could reasonably have determined that [the claimant] was not disabled before [her last insured date]. Rather, when there is no contemporaneous medical documentation, we ask whether the evidence is ambiguous regarding the possibility that the onset of her disability occurred before the expiration of her insured status. If the medical evidence is ambiguous and a retroactive inference is necessary, SSR 83-20 requires the ALJ to call upon the services of a medical advisor to insure that the determination of onset is based upon a “legitimate medical basis.”
Grebenick v. Chater, 121 F.3d 1193, 1200-1201 (8th Cir.1997). Additionally, we note that when an onset date is ambiguous, it is not usually possible for an ALJ to make a decision that is supported by substantial evidence. Finally, the Commissioner’s position ignores the possibility that a medical advisor’s assistance could change the ALJ’s determination at step three (the listed impairment step) of the five-step inquiry, thereby obviating any need to examine residual functional capacity at step five.
In contrast to the Commissioner’s argument, our precedent clearly establishes that where “medical evidence of onset is ambiguous,” an ALJ is obligated to call upon the services of a medical advisor. Reid v. Chater, 71 F.3d 372, 374 (10th Cir.1995); see also Grebenick, 121 F.3d at 1201 (“If the medical evidence is ambiguous, and a retroactive inference is necessary, SSR 83-20 requires the ALJ to call upon the services of a medical advisor to insure that the determination of onset is based upon a ‘legitimate medical basis.’ ”); see also Bailey v. Chater, 68 F.3d 75, 79 (4th Cir.1995) (“[I]f the evidence of onset is ambiguous, the ALJ must procure the assistance of a medical advisor in order to render the informed judgment that the Ruling requires.”). Proceedings before an ALJ are not inquisitorial; rather, “[a]n ALJ [has] a duty to develop a full and fair record, and therefore must consult a medical advisor” when evidence of onset is ambiguous. See Henderson v. Apfel, 179 F.3d 507, 513 (7th Cir.1999) (internal quotation marks omitted). “In the absence of clear evidence documenting the progression of [the claimant’s] condition, the ALJ *912[does] not have the discretion to forgo consultation with a medical advisor.” Bailey, 68 F.3d at 79.
Thus, the issue of whether the ALJ erred by failing to call a medical advisor turns on whether the evidence concerning the onset of Mr. Blea’s disabilities was ambiguous, or alternatively, whether the medical evidence clearly documented the progression of his conditions. Below, we examine the medical record with respect to both of Mr. Blea’s disabilities. We conclude that Mr. Blea’s medical record was ambiguous, did not clearly document the progression of his impairments, and presented a situation where the ALJ needed to infer an onset date both for Mr. Blea’s dysthymia and post-traumatic arthritis. Therefore, we remand this case to the ALJ with instructions to follow the provisions of SSR 83-20 and call a medical advisor.2
a. Posh-Traumatic Arthritis
The ALJ found that Mr. Blea did not prove he suffered from post-traumatic arthritis prior to December 31, 1998, relying primarily on three inferences he made from the medical evidence in the record. First, at a March 1998 visit with a specialist, Mr. Blea indicated that he was not interested in further surgery. “It is reasonable to infer then, that he was at least partially satisfied with the result of the surgery he had before.” Aplt’s App. vol. 1, at 22. Second, with respect to Mr. Blea’s June 1998 visit to a doctor for treatment after stubbing his toe while carrying his child, the ALJ determined that “it is apparent that at that time, he was able to ambulate effectively enough to feel comfortable carrying his baby.” Id. Third, with respect to the gap in medical evidence from June 1998 to January 2000, the ALJ stated that “[i]t stands to reason that if the claimant’s pain was so severe as to be completely debilitating as alleged, he would have sought medical attention throughout this period.” Id.
These inferences are not reasonable because they do not necessarily flow from the facts. First, it is equally reasonable to think that Mr. Blea declined further surgery because he did not wish to endure another painful process—an eighth surgery—as it is to think that he was not interested in surgery because he was fully recovered. Second, although he carried his child, he may have done so out of necessity—not because he was fully recovered. Third, as other evidence suggested, it is also reasonable to infer that the gap in medical treatment occurred, not because Mr. Blea felt better, but because he was dysthymic and dependent on alcohol.
Thus, we find these inferences insufficient to support the ALJ’s decision in light of the requirements of SSR 83-20. Mr. Blea’s medical record is indisputably incomplete during a pertinent time period, *913June to December 1998. But, rather than “calling] on the services of a medical ad-visor when onset must be inferred,” the ALJ made negative inferences against Mr. Blea due to the gap in the medical record. SSR 83-20 at 3. An ALJ may not make negative inferences from an ambiguous record; rather, it must call a medical ad-visor pursuant to SSR 83-20. Reid, 71 F.3d at 374. “The requirement that, in all but the most plain cases, a medical advisor be consulted prior to inferring an onset date is merely a variation on the most pervasive theme in administrative law— that substantial evidence support an agency’s decisions.” Bailey, 68 F.3d at 80.
In this regard, we find the Fourth Circuit’s opinion in Bailey v. Chater, 68 F.3d 75, particularly instructive. In Bailey, a claimant had a number of ailments that he acquired over a few years, including diabetes, anxiety, high blood pressure, and allergies. The ALJ had fixed the claimant’s onset date two years after the date on which she had asserted her disability began. The Fourth Circuit reversed and remanded with instructions for the ALJ to call a medical advisor pursuant to SSR 83-20 because the “evidence regarding the onset date is ambiguous.... In the absence of clear evidence documenting the progression of [the claimant’s] condition, the ALJ did not have the discretion to forgo consultation with a medical advisor.” Id. at 79.
As in Bailey, the ALJ lacked “clear evidence documenting the progression” of Mr. Blea’s post-traumatic arthritis. Id. In June 1998, six months before his last-insured date, Mr. Blea appeared not to be experiencing a significant amount of pain; however, by January 2000, approximately one year after his last-insured date, Mr. Blea exhibited symptoms, including complaints of pain, difficulty walking, and degenerative changes present on x-rays. The condition was advanced enough by February 2000 to permit a treating physician to diagnose him with “significant post-traumatic arthritis.” Aplt’s App. vol. I, at 150. With respect to the critical time, June 1998 to December 1998, however, the medical record before the ALJ was silent. Therefore, the ALJ should have called a medical advisor to assist in making reasonable inferences.
We note that on appeal, we have the benefit of a letter from Dr. Schenck, one of Mr. Blea’s treating physicians. See O’Dell, 44 F.3d at 859 (holding that any new evidence submitted to the Appeals Council on review “becomes part of the administrative record to be considered when evaluating the Secretary’s decision for substantial evidence.”) In this letter, which was provided only to the Appeals Council, not to the ALJ, Dr. Schenck opines that “it is reasonable to assume that Mr. Blea’s post-traumatic arthritis with constant pain was present between June 1998 and January 2000.” Aplt’s App. vol. I at 309. This letter further undermines the ALJ’s decision because “[a]n ALJ is required to give controlling weight to a treating physician’s well-supported opinion, so long as it is not inconsistent with other substantial evidence in the record.” Drapeau v. Massanari, 255 F.3d 1211, 1213 (10th Cir.2001). On remand, the ALJ must address Dr. Schenck’s opinion, as it is now part of the medical record.
b. Dysthymia
The ALJ declined to consider Mr. Blea’s claim that his dysthymia disabled him as of June 1997, noting:
a review of the medical record failed to provide any documentation of alleged depression or treatment for depression before Dr. Padilla’s evaluation of the claimant on August 28, 2002. Mr. Blea reported long-standing depression in *914that examination interview but the record did not support the claimant’s subjective report. Due to the complete lack of treatment or even the report of depression prior to December 1998, this impairment is not considered prior to the claimant’s date last insured.
Aplt’s App. vol. I, at 21. Mr. Blea contends that the ALJ’s findings are not supported by substantial evidence because Dr. Padilla’s evaluation specifically noted that Mr. Blea had “been dysthymic for years.” Id. at 246. Additionally, he argues that the ALJ should have called a medical ad-visor to “further address[] the onset of claimant’s depression.” Aplt’s Br. at 16. The Commissioner contends that the ALJ correctly disposed of Mr. Blea’s dysthymia claim because Mr. Blea failed to provide medical evidence of his depression prior to his last insured date.
The Commissioner’s argument and the ALJ’s decision both contradict the clear dictates of SSR 83-20, which specifically provides, as an example of when an ALJ should call a medical advisor, the situation where “the onset of a disabling impairment[] occurred some time prior to the date of the first recorded medical examination.” SSR 83-20 at 3. Additionally, we note that the ALJ’s reasoning is at tension with the Commissioner’s earlier determination that there was sufficient medical evidence for the Commissioner to determine that, as of at least March 1, 2002, Mr. Blea’s dysthymia was a “medically determinable impairment [] that [was] disabling,” a date prior to any treatment or report of depression. Aplt’s App. vol. I, at 18.
As we described above, the ALJ “should call on the services of a medical advisor when onset must be inferred.” SSR 83-20, at 3. The ALJ erred by failing to call a medical advisor “in the absence of clear evidence documenting the progression of [the claimant’s] condition.” Bailey, 68 F.3d at 79. Therefore, on remand, the ALJ should apply SSR 83-20 and call a medical advisor with respect to Mr. Blea’s claim of dysthymia as well.
B. Failure to Consider Lay — Witness Testimony
We next examine Mr. Blea’s contention that the ALJ’s decision requires remand because of the ALJ’s failure to discuss or consider the lay testimony of his wife. Becky Blea, Mr. Blea’s wife since 1994, testified before the ALJ about Mr. Blea’s depression, as well as his problems with his foot that developed after the gunshot wound. Mrs. Blea stated that her husband began expressing suicidal thoughts while he was hospitalized after the gunshot incident, and continued to talk about suicide regularly thereafter. Aplt’s App. vol. I, at 332-33. She also testified about the physical limitations her husband faced as a result of his injury: since 1997, her husband had only rarely left the house with her and their son; he has not shopped, cooked, or cleaned since then; and he cannot sleep through the night.
The ALJ’s written decision fails to mention any of the particulars of Mrs. Blea’s testimony, and in fact, never even mentions the fact that she did testify regarding the nature and severity of her husband’s impairments. Mr. Blea contends that the ALJ committed reversible error by failing to mention or discuss his wife’s testimony corroborating his account of the progression of his post-traumatic arthritis and depression. The Commissioner replies that the ALJ’s omissions are not reversible error because “the ALJ is not required to make a written finding about each witness’s credibility.” Aple’s Br. at 16 (citing Adams v. Chater, 93 F.3d 712, 715 (10th Cir.1996)).
*915The Commissioner is incorrect by only stating part of the rule in this circuit. In actuality, the ALJ is not required to make specific written findings of credibility only if “the written decision reflects that the ALJ considered the testimony.” Adams, 93 F.3d at 715. “[I]n addition to discussing the evidence supporting his decision, the ALJ also must discuss the uncontroverted evidence he chooses not to rely upon, as well as significantly probative evidence he rejects.” Clifton v. Chater, 79 F.3d 1007,1009 (10th Cir.1996).
Here, the ALJ made no mention of Mrs. Blea’s testimony, nor did he refer to the substance of her testimony anywhere in the written decision. Thus, it is not at all “clear that the ALJ considered [Mrs. Blea’s] testimony in making his decision.” Adams, 93 F.3d at 715. Additionally, Mrs. Blea’s testimony regarding her husband’s suicidal thoughts is not only uncontroverted; it serves to corroborate Dr. Padilla’s psychiatric examination of Mr. Blea, where he stated that Mr. Blea has been dysthymic for years. See Aplt’s App. at 243-46. Thus, the ALJ’s refusal to discuss why he rejected her testimony violates our court’s precedent, and requires remand for the ALJ to incorporate Mrs. Blea’s testimony into his decision. “Without the benefit of the ALJ’s findings supported by the weighing of this relevant evidence, we cannot determine whether his conclusion^] ... [are] supported by substantial evidence.” Threet, 353 F.3d at 1190; see also Baker v. Boxeen, 886 F.2d 289, 291 (10th Cir.1989) (“[W]here the record on appeal is unclear as to whether the ALJ applied the appropriate standard by considering all the evidence before him, the proper remedy is reversal and remand.”).
Additionally, Mr. Blea contends that the failure to include discussion of Mrs. Blea’s testimony contradicts two regulations: (1) SSR 83-20, which directs an ALJ to assess how lay evidence impacts the determination of an onset date; and (2) SSR 85-16, which relates only to mental impairments, and instructs that “relevant, reliable information, obtained from ... family members ... may be valuable in assessing” whether a mental impairment is disabling. SSR 85-16, at 4. The Commissioner argues that neither of these regulations is applicable to Mr. Blea’s claim.
We agree with the Commissioner that neither of these regulations necessarily apply to Mr. Blea’s claim before the ALJ, but caution that the ALJ should nevertheless be mindful of the regulations on remand. SSR 83-20 instructs an ALJ to consider “other sources of documentation” such as testimony from family when “reasonable inferences about the progression of the impairment cannot be made on the basis of the evidence in file and additional relevant medical evidence is not available.” SSR 83-20, at 3. Depending upon the “reasonable inferences” that the ALJ makes in consultation with a medical advisor on remand, it may be necessary to consider Mrs. Blea’s testimony pursuant to this provision of SSR 83-20.
With regard to SSR 85-16, we similarly believe it is premature to address whether that regulation applies. SSR 85-16 concerns “the issues to be considered when an individual with a mental impairment requires an assessment of the residual functional capacity (RFC) in order to determine the individual’s capacity to engage in basic work-related activities.” SSR 85-16, at 1. Whether the ALJ will need to follow SSR 85-16 on remand depends upon whether the ALJ’s analysis of Mr. Blea’s mental impairments reaches steps four or five of the five-step sequential analysis. See id. at 2 (explaining that an RFC assessment is unnecessary if an impairment is found to be of listing severity at step three because “[f]or impairments of listing *916severity, inability to perform substantial gainful activity (SGA) is presumed from prescribed findings”).
III. CONCLUSION
We believe it is unclear when Mr. Blea’s post-traumatic arthritis and dysthymia became disabling. Therefore, we reverse the magistrate judge’s decision and direct that this case be remanded to the ALJ for further proceedings consistent with this opinion.
. At oral argument, the attorney representing the Commissioner contended that SSR 83-20 did not apply to Mr. Blea's case because the plain wording of SSR 83-20 stated that a medical advisor was only required if the onset of a disability occurred prior to the date of the first recorded medical examination. As the Commissioner is well aware, we do not generally address arguments made for the first time at oral argument. United States v. Gonzalez-Coronado, 419 F.3d 1090, 1094 n. 7 (10th Cir.2005). We further note, however, that *910this argument misconstrues SSR 83-20, which, as we explain above, requires the assistance of a medical advisor whenever ''onset” must be inferred.
. We need not scrutinize Mr. Blea’s ancillary argument that the Appeals Council erred by not analyzing new evidence from one of Mr. Blea’s treating physicians. In denying review of the ALJ's decision, the Appeals Council stated "we considered the additional evidence listed.... We found that this information does not provide a basis for changing the Administrative Law Judge’s decision.” Aplt’s App. vol. I at 6-7. The Social Security Administration’s regulations ”require[] the Appeals Council to consider evidence submitted with a request for review.” Threet v. Barnhart, 353 F.3d 1185, 1191 (10th Cir.2003). However, where as here, the Appeals Council states that it has considered the additional evidence, "our general practice, which we see no reason to depart from here, is to take a lower tribunal at its word when it declares that it has considered a matter.” Hackett v. Barnhart, 395 F.3d 1168, 1173 (10th Cir. 2005). Because we remand this case to the ALJ, we need not determine whether the medical evidence here is so strong as to require a change in our general practice.