A common assumption is that protracted alcohol consumption is causally related to the intellectual impairment seen in sober recovering alcoholics. However, attempts to demonstrate consistent and significant relationships between alcohol intake and cognitive-perceptual functioning in such alcoholics have been inconclusive. A review of the human studies indicates that the inconsistent results cannot be accounted for by potentially confounding variables such as age, education, sex, days abstinent, years of abuse, type of hospital and number of tests used. In detoxified mice and rats there is clear evidence for a causal relationship between duration of alcohol intake and learning and memory tasks, although limited mainly to studies of relatively high levels of alcohol intake and performance on a few types of tasks. These studies suggest a duration threshold for impairment before which impairment does not occur and after which there is a systematic inverse relationship between duration and quality of performance. Finally there is an asymptote effect where, with increase in duration and increase in age, there is no corresponding decrement in performance. A number of promising areas to be investigated using the rodent model are discussed. Findings from such investigations should have heuristic value for human studies. Reasons for the inconsistent results in humans are detailed and suggestions offered to facilitate further research in this area.