Previously, we found that impaired cognition predicted treatment dropout from cognitive behavioral therapy (CBT) in a small sample of cocaine dependent patients. To further address the role of impaired cognition in retention and treatment outcome of cocaine-dependent patients in CBT, we expanded a previous investigation to a larger sample, added depressed cocaine patients, and added an additional cognitive assessment. Fifty-six cocaine dependent patients receiving CBT in outpatient clinical trials were assessed for cognitive performance at treatment entry with the computerized MicroCog (MC) and the Wisconsin Card Sort Test (WCST). Treatment completion was defined as 12 or more weeks. Treatment dropouts had significantly lower MC scores (poorer cognitive functioning) than completers on attention, memory, spatial ability, speed, accuracy, global functioning, and cognitive proficiency, with effect sizes in the moderate to large range. These findings were not affected by depression, demographics (age, gender, race, sex, marital status) or drug use (years of cocaine use or average weekly cocaine expenditure in the prior 30 days). In contrast, patients' performance on the WCST was in the average or near-average range, and WCST scores did not differentiate between completers and dropouts. Consistent with previous research, results suggest that mild cognitive impairments (< or =1 S.D. below the mean) negatively affect retention in outpatient CBT treatment for cocaine dependence. Future studies should examine whether there are specific effects of different executive functioning abilities on treatment outcome. Modified behavioral and pharmacologic interventions should be considered to target mild cognitive impairments to improve substance treatment outcome.