The effects of prenatal alcohol exposure on learning in adult offspring were studied in a spatial task in a T-maze. Male and female Long-Evans rats were selected from litters whose dams had received one of three treatments: alcohol in a liquid diet (35% ethanol-derived calories, 35% EDC), pair-fed nutritional control (0% ethanol-derived calories, 0% EDC) or standard control (lab chow, LC). The task included trial-independent (reference memory) and trial-dependent (working memory) components: subjects were required to make a fixed left-right discrimination, and then to alternate left and right choices to escape water. Prenatal alcohol exposure was associated with a greater number of reference errors for both sexes; only males from the alcohol prenatal treatment group, however, were impaired on the working memory component. These results suggest that prenatal exposure to alcohol can cause behavioral dysfunctions that persist into adulthood. Secondly, the pattern of memory impairments suggests that both sexes may be equivalently damaged in neural areas subserving reference memory, but that males are selectively more vulnerable in neural areas subserving working memory.