Angela Farrell suffers from anxiety, depression, suicidal tendencies, insomnia, vertigo, migraine headaches, fibromyalgia, carpal tunnel syndrome, and plantar fasciitis. Citing this array of impairments, she applied for disability insurance benefits; as of the date of her application, she was almost 34 years old. Her initial application was denied, but the Social Security Administration Appeals Council remanded her case for reconsideration. On remand, the Administrative Law Judge (ALJ) once again ruled against her, in part because of her failure to establish definitively that she suffered from fibromyalgia. The Appeals Council summarily affirmed this decision, despite new evidence before it that confirmed the fibromyalgia. The district court in turn affirmed that ruling, and Farrell now appeals. We reverse. The Social Security Administration’s own regulations require the Appeals Council to consider “new and material evidence,” but it did not do so in this case. In addition, several other aspects of the ALJ’s decision independently require correction. Because these warrant reversal in and of themselves (that is, without regard to the error committed by the Appeals Council), we follow the procedure that normally applies when the Appeals Council denies review and remand to the ALJ.

I

Farrell is married and has two children. She is 4'11" tall and, at the time of the hearing, she weighed 211 pounds; this represents a body mass index of 42.6, well into the range of obesity (which is 30 or greater). See NIH, National Heart Lung and Blood Institute, http://www. nhlbisupport.com/bmi/bminojs.htm. She completed between two and three years of college and has worked in a variety of jobs, including as a tax analyst, an accounting clerk, and a waitress.

Her primary physician is Dr. Sara Beyer, who has been treating her since at least 2002. According to Dr. Beyer’s reports, Farrell has suffered from progressively[*769] worsening physical and mental conditions. In 2003, Dr. Beyer treated her for panic attacks. Dr. Beyer reported that Farrell’s medications were ineffective in quelling these attacks, and she noted that Farrell was suffering from severe pain throughout her body, increased anxiety, and suicidal thoughts. Following surgery in April 2008, Farrell returned to work, but she quickly became fatigued and anxious. In response, Dr. Beyer specifically instructed her to avoid stressful situations — advice that in Farrell’s case covered a lot of ground. Practically, in order to comply she would have needed to avoid any contact with the outside world, given her photo- and phonophobias. In July of that year, Farrell underwent a psychiatric assessment in which she received a Global Assessment of Functioning (GAF) score of 51 — a score that is right on the border between “severe” and “moderate” symptoms. (A GAF score of 41-50 indicates serious symptoms; a score of 51-60 indicates moderate symptoms; and a score in the range of 61-70 indicates mild symptoms. Am. Psychiatric Ass’n, Diagnosis and Statistical Manual of Mental Disorders 32-34 (4th ed. 2000).) As the year progressed, so did Farrell’s symptoms. Her joint and back pain became worse, and her mental symptoms began to include paranoia, occasional hallucinations, nightmares, and more serious thoughts of suicide (including a specific plan to overdose on drugs).

In June 2004, Farrell’s GAF score plummeted to 30, well below the “serious” point. With new treatment for her migraines and carpal tunnel syndrome, as well as stronger medication for her depression and anxiety, her GAF score improved to 50 by September of that year, but her symptoms were still significant. She reported suffering from extreme stress in social situations, an inability to concentrate, and continuing back and joint pain.

In April 2005, Farrell complained of a constant sense of worriment and problems concentrating, as well as several new physical ailments, including an irregular heartbeat. After an examination, Dr. Beyer recorded that Farrell suffered from anxiety, insomnia, depression, joint pain, and anemia. She also alluded to the possibility of fibromyalgia — a diagnosis that both Dr. Beyer and other treating specialists had considered in the past.

Shortly thereafter, in May 2005, Farrell applied for disability insurance benefits from the Social Security Administration, alleging an onset date in November 2003. Her application noted her history of “depression, anxiety, phobias, migraines, su[icid]al tendencies, vertigo, fibromyalgia, carpal tunnel [syndrome], insomnia, [and] plantar fasciitis,” and claimed that she is unable to work as a result of these ailments.

As part of the process of evaluating her application, Farrell’s file was reviewed by several physicians engaged by the state. In general, they had a more optimistic assessment of her capabilities than her treating physicians had reached. Dr. Perkins, for example, suggested that Farrell suffered from only “moderate difficulties” and suggested that she could hold jobs requiring simple and routine tasks. After reviewing Farrell’s medical records, Dr. Pyle determined that she was able to lift and carry between 25 and 50 pounds, and work for 6 hours in an 8 hour workday. Dr. Mann thought that Farrell had only “mild restrictions in daily activities” and found her capable of jobs involving only simple tasks. Dr. Boyce, who testified before the ALJ, stated that there was no evidence in Farrell’s records of inflammation that would give rise to arthritic pain. He further testified that there was no[*770] evidence of a confirmed diagnosis of fibromyalgia.

After weighing Dr. Beyer’s conclusions against the opinions offered by the state’s reviewing physicians, the ALJ ruled against Farrell. He found the testimony of the reviewing physicians to be more consistent with the medical record, and he credited Dr. Boyce’s view regarding fibromyalgia (¿a, that she suffered from only “minimal functional limitations resulting from [her] ... fibromyalgia-type illness”) while chiding Dr. Beyer for lacking a clinical basis for her evaluation of Farrell’s functional capacity. Farrell sought review at the Appeals Council. She included new evidence with her submission, but the Appeals Council nevertheless summarily denied her petition. The district court affirmed, and this appeal now follows.

II

On appeal, Farrell presents a variety of challenges to the Social Security Administration’s decision to deny her application for disability benefits. We review the ALJ’s decision “to deny benefits to determine whether it was supported by substantial evidence or is the result of an error of law.” Rice v. Barnhart, 384 F.3d 363, 369 (7th Cir.2004).

A

Farrell’s first argument is that the district court and the Appeals Council erred by refusing to consider her new evidence confirming a diagnosis of fibromyalgia. As we noted above, the ALJ found “no evidence that this diagnosis ha[d] been confirmed” and accordingly ruled that Farrell’s claimed impairments were “non-severe.” In response to this adverse ruling, Farrell received confirmation of the diagnosis that had been suggested several times in her medical reports: tests conducted by Dr. Ryan Loyd showed that all 18 fibromyalgia points were tender, and although only 11 positive results are required for a confirmed diagnosis, Farrell tested positive in 16, including several on her neck, shoulders, knees, elbows, and chest. (The NIH’s website explains that “[t]o be diagnosed with fibromyalgia, you must have had at least 3 months of widespread pain, and pain and tenderness in at least 11 of 18 areas,” including arms (elbows), buttocks, chest, knees, lower back, neck, rib cage, shoulders, and thighs. See http://www.ncbi.nlm.nih.gov/ pubmedhealth/PMH0001463/.)

In light of Dr. Loyd’s firm diagnosis, Farrell sought review of the ALJ’s decision at the Appeals Council and included this new evidence with her application. The Appeals Council summarily denied the application. In affirming that ruling, the district court specifically refused to consider Farrell’s new evidence, citing Rice v. Barnhart, 384 F.3d at 366 n. 2, in which we ruled that it was “not appropriate for us to consider evidence which was not before the ALJ, but which [plaintiff] later submitted to the Appeals Council” because “the Appeals Council eventually refused [plaintiffs] request to review the ALJ’s unfavorable decision.” The district court was correct in its ruling. This is because 42 U.S.C. § 405(g) provides that a reviewing court may consider additional evidence “only upon a showing that there is new evidence which is material.” The evidence Farrell wanted the court to consider was not “new” to the district court because it had been already been submitted to, and rejected by, the Appeals Council. Evidence that has been rejected by the Appeals Council cannot be considered to reevaluate the ALJ’s factual findings.

Nevertheless, whether the ALJ’s decision is supported by substantial evidence is not the same question as whether the Ap[*771] peals Council properly rejected Farrell’s appeal. The Social Security Administration regulations require that body to evaluate “new and material evidence” in determining whether a case qualifies for review. 20 C.F.R. §§ 404.970(b), 416.1470. In Perkins v. Chater, 107 F.3d 1290 (7th Cir.1997), we held that “[o]ur review of the question whether the [Appeals] Council made an error of law in applying this regulation is de novo.... In the absence of any such error, however, the Council’s decision whether to review is discretionary and unreviewable.” Id. at 1294. Here, the Appeals Council’s decision says that it “considered ... the additional evidence ... [and] found that this information does not provide a basis for changing the Administrative Law Judge’s decision.”

We note that this text, which often appears in orders of the Appeals Council rejecting plenary review, is not as clear as it might be. On the one hand, it might indicate that the Appeals Council found the proffered new evidence to be immaterial, but on the other hand it might indicate that the Council accepted the evidence as material but found it insufficient to require a different result. This ambiguity is reflected in several decisions from reviewing courts. See, e.g., Brewes v. Commissioner of Soc. Sec. Admin., 682 F.3d 1157, 1162-63 (9th Cir.2012) (avoiding the question by holding, in tension with this court’s Rice decision, that the new evidence becomes part of the administrative record for purposes of “reviewing the Commissioner’s final decision for substantial evidence”); Meyer v. Astrue, 662 F.3d 700, 705-06 (4th Cir.2011) (same); Krauser v. Astrue, 638 F.3d 1324, 1328 (10th Cir.2011) (discussing ambiguity and holding that where the Appeals Council rejects new evidence as non-qualifying and claimant challenges that ruling on judicial review, that the “general rule of de novo review permits [the court] to resolve the matter and remand if the Appeals Council erroneously rejected the evidence.”). Krauser is most consistent with our ruling in Rice. See also Bergmann v. Apfel, 207 F.3d 1065, 1069-70 (8th Cir.2000); Aulston v. Astrue, 277 Fed.Appx. 663, 664 (8th Cir.2008). We thus interpret the Appeals Council decision as stating that it has rejected Farrell’s new evidence as non-qualifying under the regulation and proceed along the lines we indicated in Perkins to review that limited question.

We find the Appeals Council’s determination that Farrell’s evidence was not “new and material” to be erroneous. It is undisputed that Dr. Loyd’s diagnosis was “new” to the administrative record at the time of Farrell’s application to the Appeals Council. Its materiality is also, in our view, beyond question: the ALJ’s decision unequivocally rests in part on the determination that “there is no evidence that [a fibromyalgia] diagnosis has been confirmed.” Farrell’s new evidence fills in that evidentiary gap by providing exactly that confirmation. And this diagnosis, confirmed in December 2008, “relates to the period on or before the date of the administrative law judge hearing decision” (November 2008) as required by 20 C.F.R. § 404.970(b). It builds on the allusions to possible fibromyalgia in Dr. Beyer’s reports from 2005 and 2006. Dr. Loyd’s diagnosis was “new and material” evidence that the Appeals Council improperly failed to consider.

The Commissioner contends that “[b]e-cause the Appeals Council did not make any finding with regard to the materiality of the evidence Farrell submitted ... there is nothing in the Appeals Council’s denial of review upon which Farrell can properly pin an assertion of legal error.” This position is inconsistent with our decision in Perkins and the other decisions we[*772] discussed above; it would make the right recognized in the regulations to submit new evidence to the Appeals Council meaningless. We conclude that the Appeals Council committed legal error by ignoring Dr. Loyd’s opinion in its decision to reject Farrell’s appeal. See Scivally v. Sullivan, 966 F.2d 1070, 1075 (7th Cir.1992).

This error was not harmless. The Commissioner suggests now that the ALJ’s determination about the severity of Farrell’s fibromyalgia is irrelevant, because once an ALJ finds any severe impairment, her determination regarding the severity of the other impairments is immaterial. This is true only insofar as the severity finding relates to meeting the required threshold in step two of the ALJ’s five-step analysis. 20 C.F.R. § 404.1520(a)(4)(h) (requiring a “severe” impairment to move on to step three); Castile v. Astrue, 617 F.3d 923, 926-27 (7th Cir.2010) (“[T]he step two determination of severity is merely a threshold requirement.” (quotation marks omitted)). This is not the only place, however, in which the severity of an applicant’s conditions is properly part of the ALJ’s analysis. It also affects the ALJ’s determination of residual functional capacity, for example, and thus, no matter what happens at step two, a correct assessment remains important. See Castile, 617 F.3d at 926-27; Arnett v. Astrue, 676 F.3d 586, 592 (7th Cir.2012).

B

Farrell also challenges a number of the ALJ’s factual determinations. Many of them, such as the ALJ’s assessment of Farrell’s credibility, are supported by substantial evidence. Nevertheless, as we explain in more detail below, the ALJ failed to grapple properly with the competing medical opinions.

Farrell contends that the ALJ’s Residual Functional Capacity (RFC) determination, see 20 C.F.R. § 404.1520(a)(4)(v), improperly discounted the medical opinions of her treating physician, Dr. Beyer. See 20 C.F.R. § 404.1527(e)(l)-(c)(2) (“Generally, we give more weight to opinions from your treating sources.... ”). In response, the Commissioner argues that it is required to defer only if the treating physician’s opinion is “supported by objective clinical findings.” Henderson v. Apfel, 179 F.3d 507, 514 (7th Cir.1999). Citing to only a handful of pages in the record (and only one page from Dr. Beyer’s notes), the Commissioner suggests that no such clinical evidence exists. Dr. Beyer’s notes, however, span many years and consume many pages. We do not know what the ALJ thought about most of this material, because he never seriously discussed it.

Dr. Beyer suggested that Farrell was capable at most of only occasionally lifting over 20 pounds and that she could not sit or stand for more than 30 minutes at a time. The lifting restriction is supported by the June 2004 report, which indicates that Farrell has “severe r[igh]t index pain — hard to bend finger.” The August 2004 report similarly describes Farrell’s carpal tunnel syndrome, which requires her to “wear her wrist splints at all times.” Dr. Beyer also reported that Farrell was limited by her chronic fatigue, and that her inability to concentrate would affect her capacity to listen. The ALJ faulted Dr. Beyer for “not referencing] clinical evidence to support [these] proposed restrictions.” But a careful examination of the record that Dr. Beyer furnished shows exactly the kind of supporting evidence the ALJ apparently wanted. As early as May 2002, Dr. Beyer noted that Farrell suffered from “severe fatigue” and that it was “hard [for Farrell] to get out of bed.” In[*773] September 2002, Farrell complained of “being tired all the time” (emphasis in original). In 2003, Dr. Beyer noted that Farrell braises easily (incidently, a point not dependent on Farrell’s self-reporting), and later that year diagnosed her with restless leg syndrome (a neurological condition that is similarly not dependent on subjective reports). Beginning in 2004, Dr. Beyer’s notes make increasing references to joint pain and lower back pain.

Notably, just before Farrell’s alleged onset of disability — in October 2003 — Dr. Beyer suggested that the “real cause” for Farrell’s “severe fatigue” was her “stress [and] depression.” Similarly, in September 2004, Dr. Beyer suggested that Farrell’s “concentration problems” might stem from her “depression [and] anxiety.” The ALJ’s decision makes almost no mention of these mental ailments — diseases which are best evaluated by those physicians who have a long history of treating the applicant — despite the fact that Dr. Beyer’s medical reports repeatedly suggest that these mental conditions may be the root cause of some of her physical limitations.

As we already have pointed out, Dr. Beyer’s reports indicate that Farrell was suffering from anxiety and depression as early as 2003. For example, her notes from July 2003 state that Farrell was suffering from “increased anxiety ... Reading] her to increased panic attacks,” and the report from December flags increasing anxiety. These ailments persisted. Dr. Beyer’s reports from 2005 and 2006 continue to refer to Farrell’s anxiety, stress, depression, along with new bouts of panic attacks in June 2005. Farrell was hospitalized on more than one occasion because of her suicidal tendencies, and she admitted to cutting her wrist with a plastic knife to relieve stress. Dr. Beyer noted that Farrell found it difficult to let go of situations that were beyond her control. It was only the thought of her own children that deterred her from committing suicide. It is true that some reports, such as those from August 2004, indicated improvements in Farrell’s condition, but these successes were only temporary. Farrell’s GAF score similarly vacillated, but it only sporadically moved outside of the “severe” zone. Farrell’s RFC should not have been measured exclusively by her best days; when a patient like Farrell is only unpredictably able to function in a normal work environment, the resulting intermittent attendance normally precludes the possibility of holding down a steady job. Cf. EEOC v. Yellow Freight Sys., 253 F.3d 943, 949-52 (7th Cir.2001) (en banc). Matters would be different if the ALJ had confronted Dr. Beyer’s opinions and had explained why he was rejecting them. But he did not. Instead, he ignored the extensive medical history in the record and emphasized contradictions with the opinions of the government’s doctors. This was error. See Gudgel v. Barnhart, 345 F.3d 467, 470 (7th Cir.2003).

Farrell also challenges the hypothetical questions that the ALJ posed to the testifying vocational experts, alleging that they did not fully incorporate his findings regarding her RFC. Because we have determined that the ALJ’s RFC determination is based on an incomplete assessment of the record and does not account for Dr. Loyd’s diagnosis, we need not decide whether the ALJ’s examination was appropriate. On remand, the ALJ may need to re-examine these or other experts in order to assess Farrell’s ability to work in light of the fresh look at the case.

The decision of the district court is Reversed, and the case is Remanded to the agency for further proceedings consistent with this opinion.