The previously-observed attenuation of withdrawal reactions in mice (group B) fed an ethanol diet containing chlordiazepoxide (CDP) was not due to a difference in the rate of disappearance of blood ethanol levels during chronic diet treatment in group B compared to mice which received only the ethanol diet (group A). Injection of group A mice with CDP or N-demethyl CDP (10 mg/kg) at the time of diet withdrawal did not result in any significant attenuation of withdrawal scores. Injection of the lactam metabolite of CDP (LCDP; 10 mg/kg) resulted in significantly attenuated withdrawal scores at 4 and 6 hr only, but the pattern of withdrawal scores were different from that for group B mice. Moreover, blood LCDP level, in mice injected with LCDP, at 4 hr was at least five times higher than that attained in group B mice (from diet containing CDP). These results support our previous conclusion that the presence of major metabolites of CDP during withdrawal could only account for a minor contribution to the protective effect. Mice in A and B did not differ in the degree of functional tolerance which developed as a result of ethanol intake. Thus, there was an apparent dissociation between tolerance and physical dependence in the mice which had consumed the CDP/ethanol diet. The magnitude of decrease of GABA levels in the cerebellum and cerebral cortex at 4 hr after withdrawal also did not differ between the two groups, suggesting the reduction in GABA levels could not be correlated with the intensity of withdrawal signs.