ORDER
Helen Hill applied for disability insurance benefits and supplemental security income benefits, see 42 U.S.C. §§ 416, 428, 1381a, 1382c, claiming that she was disabled because she suffered from diabetes mellitus, hypertension, glaucoma, and hepatitis C. The Social Security Administration initially, upon reconsideration, and after a hearing before an administrative law judge, rejected her claims. The Appeals Council denied review. Hill sought review in the district court, which upheld the determination that Hill was not disabled. Because the ALJ’s credibility determination was problematic, he ignored important evidence, and some of his findings were not reflected in the residual functional capacity (RFC) he assigned to Hill, we vacate and remand.
Hill is 52 years old and has a tenth-grade education. In 1999 — the year she says she became disabled — she had been working as a teacher in a daycare center. She also has a history of drug abuse. In January 2000 she was treated for swallowing two rocks of crack cocaine and admitted that she had used crack for 6 years. Seven months later she went to the hospital complaining of cramping in her abdomen and admitted that on the previous night she had used cocaine. The record contains no evidence that Hill used illegal drugs after 2000.
Hill received medical care in emergency rooms; from her primary-care physicians, Dr. M. Pruitt and Dr. Avery Hayes; and later at the Bread of Healing Clinic. In July 2002 Hill visited the emergency room of St. Joseph’s Hospital in Milwaukee, Wisconsin, reporting that she felt tired, dizzy, and shaky, and that she had run out of insulin two days before because she did not have sufficient insurance to pay for it. She reported that she also experienced polydipsia (excessive thirst) and polyuria (large volume of urine). The hospital gave her insulin and discharged her. In November 2002 she went to Aurora Sinai Medical Center in Milwaukee complaining of urinary frequency and experiencing an unusual vaginal discharge with itching. Doctors treated her with antibiotics. The same month Dr. Pruitt examined her and noted that her diabetes was not under control and that she was not exercising, controlling her eating, or addressing her weight. By January 2003 Dr. Pruitt noted that Hill’s diabetes was under good control and that she was eating better and exercising, though irregularly.
In May 2003 Hill went to the emergency room because she had run out of insulin two days before, her blood-sugar level was high, and she was suffering from urinary frequency, painful urination, and excessive *79amounts of urine. Doctors diagnosed her with hyperglycemia and a urinary tract infection and gave her antibiotics and insulin. A month later, in June 2003, Hill again returned to the hospital, this time because she felt jittery. Her blood-sugar level was elevated, and records in the device she uses to monitor her diabetes revealed that her levels were often high, leading the doctor to conclude that Hill “has very poor control” of her diabetes. Hill admitted that she was not following a diabetic diet. The doctor encouraged her to change her diet, lose weight, walk every day for 20 minutes, and follow her insulin regimen. In August 2003 Hill again complained of polyuria, and the doctor observed that, despite Hill’s compliance with her insulin regimen, her diabetes was not under control.
In August and September 2003, Hill underwent laser surgery for glaucoma. In October 2003 Hill reported to Dr. Hayes that she was experiencing pain in her shoulder. Dr. Hayes ordered an X-ray and noted that her diabetes still was not well-controlled.
In November 2003 Hill was seen by a specialist in hepatology (liver function). Although in 2000 she had admitted using crack and cocaine, Hill told the heptologist that she had not taken illicit drugs since her hepatitis C diagnosis 18 years earlier. The heptologist noted that her symptoms included fatigue, weight gain, pruritis (itching) on her back, shortness of breath with approximately one block exertion, gastrointestinal problems, nausea, and frequent urination. Hill also reported experiencing “some depression and anxiety.”
Also in November 2003 Hill saw Dr. Hayes and again complained of pain in her shoulder. Dr. Hayes noted that Hill’s X-rays had revealed degenerative changes in the acromioclavicular joint. He reported grinding and crackling (crepitus) when he moved her shoulder and that her range of motion was limited because of pain. Dr. Hayes diagnosed her with arthritis, prescribed capsaicin cream, and referred her to physical therapy. He also noted diagnoses of type II diabetes and hypertension.
In December 2003 an SSA medical consultant reviewed Hill’s records and found that her claims of fatigue were credible based on her obesity (which was “just shy of morbid”), her uncontrolled diabetes, and her chronic hepatitis C. The consultant concluded that Hill had the RFC for medium work. In April 2004 a disability examiner and another medical consultant approved these findings.
In January 2004 Hill’s diabetes remained uncontrolled; thus, Dr. Hayes increased her insulin levels. In February 2004 Hill again saw Dr. Hayes and complained of pain in her left shoulder. The doctor examined her shoulder and noted that it was tender, that passive flexion was limited to 70 degrees — meaning that the doctor could move her left arm only 70 degrees in front of her — and that abduction was 50 degrees — meaning she could raise left her arm only 50 degrees out to the side. See http://www.nlm.nih.gov/ medlineplus/mplusdictionary.html (last visited Sept. 18, 2008). He noted that arthritis probably caused these restrictions in her movement. He recommended physical therapy to “improve her range of motion and prevent a frozen shoulder.”
From April 2005 until February 2006, Hill received medical care at the Bread of Healing Clinic, which offers free health care to those without insurance. See http://crossoflife.org/catrindex_93.php (last visited Sept. 18, 2008). While she was being treated at the clinic, her blood-sugar level remained high and her diabetes was not consistently under control. She often ran out of medication and sometimes re*80ported going without it for days or months at a time. Tests showed increased levels of glucose and hemoglobin A1C.
In May 2005 Hill complained of tingling in her left hand and chest pain. Her diabetes and hypertension were not under control and she began taking a new medication for diabetes. In September she reported experiencing fatigue, headaches, depression, and thoughts of suicide. She received a prescription for Zoloft to combat her feelings of depression. In October she said that her hand was numb and tingling. In December she said that her depression was worsening, and so her dosage of Zoloft was increased. The doctor also noted that her diabetes was poorly controlled. During her January 2006 visit, she reported “ ‘feeling down and out,’ ” and that she had trouble sleeping and thoughts of hurting herself. The doctor increased her dosage of Zoloft again and recommended that she speak to a counselor. The doctor noted diagnoses of hypertension, diabetes, hepatitis C, and depression. In February she again reported tingling her hands and said that her thirst and urination had increased and that she felt sluggish and tired. The doctor noted that her diabetes was still poorly controlled, despite mild improvement.
At the hearing before the ALJ, Hill testified that she suffers pain and fatigue. She reported that she experiences frequent urination and has to go to the bathroom every five minutes. During her hearing, which lasted approximately forty minutes, she took three bathroom breaks. Although urinary frequency hampers her daily activities, she is able to take public transportation and has never sought treatment specifically for the problem. She also said that her hands tingle as if there were “pins” in her fingertips, that her little finger sometimes bends over her middle finger, and that occasionally she has trouble grasping objects and holding pencils. She testified that she feels sharp pain “all the time” in her back, shoulders, knees, and ankles, and that, although medication alleviates the pain a bit, she cannot climb stairs, take long walks, or sit for long periods of time. She takes seventeen pills a day plus insulin injections, however, she takes only prescription-strength Ibuprofen to alleviate her pain.
In response to hypothetical situations posed by the ALJ, a vocational expert (VE) testified that there are over 10,000 unskilled jobs in Wisconsin — including work as a general office clerk (2,500 jobs), counter clerk (900 jobs), information clerk (1,500 jobs), and cashier (6,000 jobs) — that involve only light work, are limited to 30 minutes at a time sitting and standing and require no more than frequent bilateral reaching (reaching with both arms) in all directions. The VE further testified that Hill could not work at a daycare center with these restrictions. She further represented that there were about 4,700 sedentary jobs in Wisconsin that could be performed with the same sitting, standing, and reaching restrictions. Finally, in response to a hypothetical situation posed by Hill’s attorney, the VE said that there were no jobs available for those who can lift 10 pounds occasionally and less than 10 pounds frequently; cannot meet strict deadlines, work at a fast pace, or work in a noisy environment; and require unscheduled breaks.
The ALJ followed the five-step analysis required by 20 C.F.R. § 404.1520 and concluded that Hill was not disabled because there are jobs in Wisconsin that she could perform with her limitations. He found, first, that Hill had not participated in gainful activity since she allegedly became disabled; second, that her diabetes, hepatitis C, arthritis, heart disorder, hypertension, glaucoma, and depression are severe im*81pairments; and, third, that none of her impairments was conclusively disabling under 20 C.F.R. § 404, Subpt. P, App.l. At the fourth step the ALJ determined Hill’s RFC, concluding that she could perform unskilled, light work that does not require more than frequent reaching in all directions and allows her to change position every half hour. He then found that Hill’s RFC prevented her from working in her prior job as a daycare teacher. At the fifth step the ALJ determined that there were almost 11,000 jobs in Wisconsin that Hill could perform.
Hill argues that, in determining her RFC, the ALJ ignored evidence about her shoulder pain, urinary frequency, tingling hands, depression, and fatigue and that the ALJ’s credibility determination is baseless. The ALJ’s decision becomes the final decision of the Commissioner where, as here, the Appeals Council declines to review it. See Haynes v. Barnhart, 416 F.3d 621, 626 (7th Cir.2005). We will affirm the ALJ’s decision so long as substantial evidence supports it — that is, if the decision is based on evidence that “a reasonable mind might accept as adequate to support the conclusion.” Richardson v. Perales, 402 U.S. 389, 401, 91 S.Ct. 1420, 28 L.Ed.2d 842 (1971) (internal quotation marks and citation omitted); see Getch v. Astrue, 539 F.3d 473, 480 (7th Cir.2008). If, like here, the ALJ denies benefits, then the ALJ must “build an accurate and logical bridge from the evidence to his conclusion,” Clifford v. Apfel, 227 F.3d 863, 872 (7th Cir. 2000), supporting the decision with adequate evidence and explaining why evidence supporting the claim was unpersuasive, see Berger v. Astrue, 516 F.3d 539, 544 (7th Cir.2008). The ALJ is not required to discuss every bit of evidence in the record, see Getch, 539 F.3d at 480, but the ALJ may not ignore entire lines of evidence that conflict with his conclusion. See Golembiewski v. Barnhart, 322 F.3d 912, 917 (7th Cir.2003). Furthermore, the ALJ must not disregard subjective complaints of disabling pain merely because they are more severe than what the medical record supports. See Johnson v. Barnhart, 449 F.3d 804, 806 (7th Cir.2006). Instead, the ALJ must develop the record and seek information about the severity of the pain and its effects on the applicant. See Clifford, 227 F.3d at 871-72.
Hill’s arguments that the ALJ’s reasoning was flawed and that he gave short shrift to her evidence convince us that the ALJ’s assessment of her RFC is not based on substantial evidence. To begin with, the ALJ found that Hill “has poor credibility, as the record reflects past deception regarding drug abuse,” but we agree with Hill that this finding is problematic. We defer to the ALJ’s assessment of a witness’s credibility as long as the record contains some support for it. See Berger, 516 F.3d at 546; Sienkiewicz v. Barnhart, 409 F.3d 798, 803 (7th Cir.2005). An ALJ is entitled to view with skepticism the testimony of an applicant who has been deceptive. See Berger, 516 F.3d at 546 (ALJ permitted to make adverse credibility finding both because applicant received pain medication from two different doctors and because he lied on his tax returns). The record here contains evidence that Hill once lied to a doctor about the last time sue used illegal drugs, and the ALJ was permitted to note her deception in determining that she “has poor credibility.” It is unclear from the decision, however, what impact this determination had on the ALJ’s assessment of the evidence. Under Social Security Ruling 96-7p, an ALJ’s evaluation of a applicant’s credibility must be specific enough to make clear to us how much weight the ALJ gave to the applicant’s testimony and the reasons for that decision. See Arnold v. Barnhart, 473 F.3d 816, 822 (7th Cir.2007); Zurawski v. Halter, 245 F.3d 881, 887 (7th Cir.2001). Here, the ALJ’s statement that he dis*82trusts Hill comes in the middle of a paragraph discussing her allegations of diabetes, urinary frequency, arthritis, hepatitis, heart disease, glaucoma, and depression, but the ALJ does not say which parts of her testimony he does not believe, and what, if anything, he credits despite her predilection to understate her history of drug use. Thus, it is impossible to evaluate the credibility determination.
We are also troubled by the ALJ’s finding that Hill’s diabetes “is poorly controlled primarily because she fails to take insulin on a regular basis.” The ALJ does not say directly that this finding caused him to disbelieve Hill, but, given the ambiguous discussion of Hill’s credibility, this determination might have affected the ALJ’s view of her testimony. Although Hill’s doctors often described her diabetes as “poorly controlled,” none of them concluded that Hill’s failure to take insulin caused the problem. Indeed, her doctors often attempted to remedy the problem by adjusting her insulin regimen or by changing her medication. And even though Hill sometimes went for days or weeks without taking insulin, the record shows that she was indigent and might have had difficulty covering the costs. An ALJ can make an adverse credibility determination based on an applicant’s failure to follow a treatment regimen if the applicant does not have a good reason for the lapse, but the ALJ first must ascertain the applicant’s reason for noncompliance. See Craft v. Astrue, 589 F.3d 668, 678-79 (7th Cir.2008). And although ability to pay for treatment is relevant to the issue of credibility, see id., the ALJ did not ask Hill why she stopped taking her insulin and why she was unable to visit a clinic before she ran out.
Hill also asserts that the RFC determination did not adequately account for the arthritis in her left shoulder. The ALJ credited Hill’s medical records showing that she has been diagnosed with arthritis, acknowledging that the records contain “a little evidence of arthritis in the shoulder with pain and a limited range of motion in the shoulder.” Although records from 2004 show that Hill could not raise her left arm above shoulder level, that there was tenderness in her left shoulder, and that the shoulder produced crackling noises when moved, the ALJ concluded that Hill could perform work that requires “frequent reaching in all directions.” The ALJ faded to explain how he could credit reports that Hill was experiencing pain and tenderness and that her range of motion was limited but still conclude that she was able to reach in all directions; thus, the ALJ failed to build a logical bridge between the evidence and his findings.
The only explanation for the ALJ’s finding that Hill’s pain “is simply not that severe,” is that Hill “does not take any prescription pain medications but only Ibuprofen” to relieve it. But the ALJ misstates Hill’s testimony; she actually said that she takes prescription-strength Ibuprofen for her pain. More troubling is that the ALJ does not say how he reached his conclusion that painful arthritis necessitates stronger medication. Ibuprofen is a common treatment for arthritis. See http://www.webmd.com/osteoarthritis/ guide/osteoarthritis-basics?page=2 (last visited Sept. 18, 2008). Moreover, Hill also suffers from hepatitis C, and her medical records show that stronger anti-inflammatory medications may damage her liver. Thus, we conclude that the ALJ’s conclusion that Hill is capable of reaching in all directions is not supported by substantial evidence.
The ALJ similarly failed to link his finding about the severity of Hill’s urinary frequency with his RFC determination. The ALJ did not credit Hill’s testimony that she required bathroom breaks every five minutes because he found it “implausible” that she would not seek treatment for *83a “chronic” problem. He found, instead, that her urinary frequency was “episodic,” but he still did not include allowances for unscheduled breaks in the RFC. This omission is curious because, even if Hill experienced only episodes of urinary frequency, she would, at times, need to take unscheduled and possibly frequent bathroom breaks. But the ALJ did not make a finding about how often Hill experiences frequent urination or about how many bathroom breaks she would need during these episodes. Moreover, nothing in the record supports the finding that Hill never sought treatment for urinary frequency. As the ALJ noted, she often complained about the problem to her doctors. Frequent urination is a symptom of diabetes, see http://www.mayoclinie.com/health/type2-diabetes/DS00585/DSECTION= symptoms (last visited Sept. 18, 2008), so it is unclear what kind of treatment, outside of her normal treatment for diabetes, the ALJ expected her to seek.
Hill also contends that the ALJ erred by failing to consider evidence that she sometimes experiences tingling, numbness, and cramping in her hands. She complained to her doctors about this problem and testified that occasionally her hands cramped so much that her little finger bent over her middle finger and that sometimes she was unable to hold a pencil. She said that this condition would prevent her from working on an assembly line. Despite this evidence, the ALJ did not say whether he believed that she suffered from these problems and whether they would limit her ability to work. The ALJ’s silence is significant; if an applicant for disability benefits has difficulty using her hands to manipulate objects, there are fewer jobs available to her. See Golembiewski, 322 F.3d at 918. Here, the VE determined that 6,000 out of the approximately 11,000 jobs Hill purportedly can perform involve duties as a cashier. The VE did not identify the category number for the specific job she had in mind, but, according to the Dictionary of Occupational Titles, a cashier, among other duties, “[rjeceives cash from customers or employees,” “[mjakes change, cashes checks, and issues receipts or tickets to customers,” “[m]ay operate ticket-dispensing machine, “and” [m]ay press numeric keys of computer.” See http://www.occupationalinfo.org/21/ 211462010.html (listing job duties for Cashier (II) clerical); see also http://www. occupationalinfo.org/21/211462010.html (same for Cashier-Checker (retail trade)) (both last visited Sept. 18, 2008). All of these functions require manual dexterity, and a person who experiences tingling and numbness in her hands and who has trouble holding objects may not be able to perform them. Thus, the ALJ should have addressed the evidence that Hill’s ability to use her hands is limited. See Golembiewski, 322 F.3d at 918.
Hill next faults the ALJ for failing to take into account her depression in determining whether she had the ability to perform light work. According to Hill, the ALJ was required to calculate her mental RFC because he had labeled her depression a “severe” impairment at step two of the analysis. An ALJ must take into account an applicant’s mental limitations when determining the RFC. See 20 C.F.R. § 404.1545(c); Craft, 539 F.3d at 675-76. Here the ALJ did discuss the evidence of Hill’s depression. He pointed out that her medical records reflect her complaints of depression and show that she takes Zoloft to control it. He then observed that Hill did not mention her depression in her testimony and did not allege that it limited her ability to work. But, he did not explain how he came to the conclusion that her depression — which he found to be a “severe impairment” — has no effect on her ability to work. The ALJ’s reasoning leads us to question whether he carefully considered all of the evidence.
*84Finally, Hill asserts that the ALJ did not discuss her allegation of fatigue despite her testimony and the finding of the SSA consultant that her complaints of fatigue are credible. Although an ALJ need not discuss every piece of evidence in the record, the cursory analysis undertaken by the ALJ here, including his failure to mention Hill’s complaints of fatigue, again causes us to question whether the ALJ considered the entire record. See Arnold, 473 F.3d at 823.
The ALJ’s rationale for the RFC is insufficient to convince us that the finding is based on substantial evidence. We therefore REVERSE the judgment of the district court and REMAND to the case to the agency for further consideration, which should include reasoned assessments of Hill’s credibility and her RFC.