To determine the validity of substance-abusing (SA) patients' self-reports of cognitive impairments, we assessed the independent contributions of depression, actual neurocognitive performance and an index of cognitive decline, in predicting cognitive complaints in groups of SA patients and normal controls. The SA sample comprised 74 veterans enrolled in day treatment. The non-clinical sample consisted of 150 English-speaking adults. Assessment instruments were as follows: A modified version of the Patient's Assessment of Own Functioning Inventory (PAOFI) containing three subscale on: Memory, Language and Communication, and Higher Cognitive Functions; the Beck Depression Inventory; a battery of neuropsychological tests that measured domains of executive function, processing speed, verbal fluency and verbal and visual memory; and a measure of premorbid intellectual functioning. SA patients reported twice as many PAOFI complaints as non-clinical controls. SA patients' neuropsychological performance was lower than that of non-clinical controls. A higher percentage of SA patients had significant cognitive decline. The SA sample reported more depression. There was no association between PAOFI scores and neuropsychological performance for either group. PAOFI results were not associated with cognitive decline. BDI scores accounted for 12% of the variance in PAOFI total score for the SA sample and 44% for the non-clinical sample in multiple regression analysis. Cognitive complaints were related more to depression than cognitive performance for both SA and non-clinical samples. The results do not support self-report as a valid means of neuropsychological assessment in SA samples, although self-reports may provide other information about perceived cognitive difficulties that may be relevant to clinical evaluation.