We have examined if long-term (13 months) alcohol consumption and the same treatment followed by a 6-week withdrawal period cause different neuropathological changes in rats. Spatial reference and working memory of alcohol-consuming and withdrawn rats were evaluated by comparison of their performance with age-matched controls in the Morris water maze. In the reference memory task we did not observe significant cognitive deficits in rats continuously exposed to ethanol, whereas withdrawn animals showed an obvious impairment of their overall performance. The reference memory deficit in withdrawn rats was evident in the spatial probe trial; these animals required significantly longer swimming distances to approach the former position of the platform when compared with controls and alcohol-consuming animals. In contrast, working memory was not significantly altered in either experimental group. Stereological methods were applied to compare the neurodegenerative changes produced by alcohol intake and withdrawal in the hippocampal formation. In the alcohol-consuming animals there was a significant cell loss in CA1 (18%) and CA3 (19%) hippocampal regions. Moreover, in withdrawn rats there was a further decay in the total number of pyramidal neurons, which amounted to 15% relative to nonwithdrawn animals. In the granular layer of the dentate gyrus there was a trend in the same direction, but it did not reach significance. Thus, our findings indicate that withdrawn rats are cognitively impaired relative to animals submitted to continuous alcohol consumption and to age-matched controls, which fits the morphological data showing that withdrawal aggravates ethanol-induced degenerative processes in the hippocampal formation.