Pavlovian conditioning models have led to cue-exposure treatments for drug abuse. However, conditioned responding to drug stimuli can return (be renewed) following treatment. Animal research and a previous study of social drinkers indicated that extinction is highly context dependent but that renewal could be reduced by the inclusion of a cue from the extinction context. This study extends this research to a clinical sample. Alcohol-dependent outpatients (N = 143) completed an extinction trial to reduce craving and salivation responses to alcohol cues. They were then randomized to renewal tests in either the same context as extinction, a different context, the different context containing an extinction cue, or the different context with cue plus a manipulation to increase the salience of the cue. Contrary to predictions, the different context did not produce the expected renewal effect. Although the generalization of extinction effects beyond the cue-exposure context is a positive clinical finding, it is inconsistent with basic research findings on the context dependence of extinction. Possible explanations for this inconsistency are discussed.