One hundred alcoholic patients were followed at monthly outpatient clinics for 6 months. Half were assigned low to moderate doses of lithium carbonate and half to chlordiazepoxide (10 mg tid) (active placebo). Drinking behavior and medication compliance were monitored at monthly clinics. After 6 months 52% of the lithium and 44% of the chlordiazepoxide patients were medication compliant. Of the 48% remaining in the lithium group, 14% did not return for a single visit while 14% came only once. Twenty percent came to clinic regularly; however they had never taken the medication as assigned. Of the 56% non-compliant chlordiazepoxide patients, 16% did not attend a single clinic; 24% came only once or twice and the remaining 16% attended clinics regularly although they were never medication-compliant. Drinking days and the percentage of patients reporting abstinence for one or more months were determined for medication compliant patients and for patients who attended clinic regularly but who did not take medication. Data were analyzed by Kruskal-Wallis one way analysis of variance tests. While lithium and chlordiazepoxide compliant patients tended to report fewer mean drinking days per month (4.6 and 4.8 respectively) than the non-medication group (6.9) these differences were not significant; however, compared to 44% in the non-medication group, 60% of the lithium patients and 58% of the chlordiazepoxide patients reported having significantly more months of abstinence ( p < .05). These results do not show that lithium is differentially efficacious in reducing alcohol consumption.