Using a test of working memory which is not influenced by learning, the win-shift foraging paradigm, ethanol was found to produce amnesia in mice. This effect was produced by very low doses of ethanol, 0.5 and 1.5 g/kg, which did not alter motivation as measured by latency to leave the start box. These doses of ethanol also did not alter performance in the maze as measured by the time to traverse the maze. These data indicate that low doses of ethanol directly impair working memory-that is, the effects are not due to an alteration in learning. The win-shift foraging paradigm, which is free of some of the problems inherent in passive avoidance paradigms, may be useful in exploring the mechanism of this effect of ethanol.