OBJECTIVES: To construct the social, neurological and neuropsychological profiles of patients with suspected alcohol related brain damage, and to test the hypothesis that cognitive dysfunction develops in alcoholics before signs of alcohol related neurological disease. DESIGN: The study design comprised: a retrospective analysis of patients' records; a comparison of patients' demographic and social variables with age-matched population data from the 1986 Census; and a comparison of neuropsychological test scores of male patients with suspected alcohol related brain damage and age-matched controls. SETTING: The Neuropsychology Unit of Royal Prince Alfred Hospital, a tertiary referral teaching hospital. PARTICIPANTS: All patients (n = 641) referred to the Neuropsychology Unit for assessment of suspected alcohol related brain damage from July 1, 1987 to June 30, 1989, and a control group (n = 93) consisting of healthy volunteers who drank no more than 40 g of alcohol per day. RESULTS: Of the patients, 515 (80%) were men and 410 (64%) were 55 years or younger; their median daily alcohol consumption was 180 g. Compared with the general population the patients were significantly more likely to be in receipt of social security payments, to be separated, divorced or widowed and living in rented or temporary accommodation. However, the proportion of professional and skilled individuals was the same as in the general population. Cerebellar degeneration (38%), peripheral neuropathy (34%) and seizures (14%) were the most frequent neurological disorders. Frontal lobe dysfunction (58%) and short-term memory loss (32%) were the most frequent cognitive abnormalities. Dementia was relatively uncommon (4%). The mean age of male patients with alcohol related brain damage but no neurological disorders (43 years) was significantly less than the mean age of those with neurological disease (54 years; P less than 0.001). In the former, neuropsychological tests of frontal lobe function showed significant impairment compared with control subjects (P less than 0.001), despite normal intelligence. CONCLUSIONS: Cognitive impairment in alcoholics frequently takes the form of frontal lobe dysfunction and may be relatively subtle, requiring a neuropsychological examination for diagnosis. Signs of cognitive impairment may precede those of alcohol related neurological disorders by more than ten years.