Rhesus monkeys were trained to respond under a variable interval 2 min schedule for codeine or ethanol injections. Both codeine and ethanol were effective in the initiation of variable-interval responding; responding was maintained over a range of codeine (0.003-1.0 mg/kg/injection) and ethanol doses (32.0-560 mg/kg/injection). Maximum rates of responding were obtained at the 0.01 mg/kg/injection codeine dose (0.14 responses/sec) and at the 180 mg/kg/injection codeine dose (0.19 responses/sec). Rates of responsing were bitonic functions of the reinforcer dose for both codeine and ethanol; maximum rates were obtained at intermediate doses and lower rates occurred at the extremes of the dose range. Both codeine and ethanol showed within-session decreases in responding across the range of reinforcer doses. Codeine-reinforced responding declined in rate within the one-hour session without a similar change in the frequency of drug injection; in contrast, both ethanol-reinforced responding and the frequency of ethanol injections declined within each session across a range of doses. Increasing or decreasing the codeine dose half-way through the one-hour session resulted in increases or decreases in codeine responding compared to controls. These data indicate that the progressive decline in codeine-reinforced responding is not the result of a generalized disruption of responding.