A shadow moving overhead acts as a danger stimulus and elicits an escape response in the crab Chasmagnathus granulatus that habituates promptly and for a long period. The effect of acute ethanol treatment on this long-term memory was analyzed. A single injection of 0.01, 0.05, or 0.1 micrograms ethanol (ET)/g given 30 min before iterated presentation of a visual danger stimulus failed to affect short-term habituation. Posttraining ethanol (0.01 to 0.1 microgram/g) produces a dose-dependent impairment of long-term habituation, but pretraining ethanol had no amnestic effect. However, a retention deficit confined to context memory was disclosed with both pre- and posttraining ethanol. Results from experiments with double injection (posttraining and pretesting injections) account for the retention impairment in terms of true amnesia (failure to acquire memory) but not due to state-dependence or retrieval deficit. The nonamnestic effect of pretraining ethanol upon long-term habituation is explained by a nonespecific depressing effect caused by interaction between iterative presentation of the danger stimulus and drug-induced internal state during training.