Levels of plasma testosterone (T) and luteinizing hormone (LH) were evaluated in alcoholic and control subjects while they smelled a real beer and then drank a placebo beer which they believed contained alcohol. Plasma T and LH differed in the alcoholics and controls during beverage presentation and T levels also differed during drink consumption. Although these results suggest a role for conditioned responses, the lack of concordance between the hormone levels and psychophysiological responses or self-report desire to drink does not support a conditioning hypothesis that could account clearly for consumption behavior. Alternative hypotheses are discussed.