Four groups of males (n = 6 each) were trained on a complex psychomotor task prior to attending five drinking sessions. They performed the task under the same dose (0.84 ml absolute alcohol/kg) on sessions 1-4. On session 5 they received a placebo. During these sessions, the first group (CR) received information and a monetary reward that was contingent on the display of nonimpaired performance under alcohol. The second group (IO) only received information about performance. The third group (NCR) received noncontingent information and reward, administered on a predetermined schedule. The fourth group (NR) received no information or reward for performance. These manipulations were predicted to influence the order of groups with respect to the rate at which tolerance developed over sessions 1-4. The findings confirmed this hypothesis, showing the rate of CR greater than IO greater than NCR greater than NR. The placebo session tested for the classically conditioned compensatory response unattenuated by alcohol. The strength of the response, observed as facilitated performance, was predicted and found to relate to the rate at which tolerance had developed (i.e., CR greater than IO greater than NCR greater than NR). This evidence for operant and classical conditioning effects in alcohol tolerance was considered to suggest that a consideration of their joint effects may be required to understand the contribution of learning to drug tolerance.