Long-Evans rats (n = 12) were trained to lever-press on a fixed-ratio 4 schedule of reinforcement with ethanol (10% v/v) presented as the reinforcer. After implantation of bilateral stainless-steel guide cannulae aimed at the nucleus accumbens, site-specific microinjections of muscimol (1-30 ng) and bicuculline (1-10 ng) were tested for effects on ethanol-reinforced responding. Baseline response patterns were characterized by initial high rates that terminated abruptly after approximately 20 min. Muscimol administration in the nucleus accumbens decreased the total number of ethanol-reinforced responses and obtained reinforcers. Bicuculline also decreased ethanol-reinforced responses and reinforcers at the highest dose tested. When a dose of bicuculline (1 ng) that was ineffective by itself was coadministered with an effective dose of muscimol (10 ng), the muscimol-induced decreases in responding were blocked. Analysis of response patterns showed that muscimol decreased ethanol self-administration by terminating responding, normally lasting 20 min, after approximately 10 min with no changes in local response rate. Bicuculline decreased total responding by producing parallel, but nonsignificant, changes in time course and response rate. These data suggest that GABAergic transmission in the nucleus accumbens is involved in the termination, but not the onset or maintenance of ethanol self-administration. The specificity of this effect gives emphasis to the importance of measuring behavioral parameters, as well as products of behavior (such as intake volume) in the study of ethanol self-administration.