Behavioral and pharmacotherapeutic approaches constitute two prominent strategies for treating cocaine dependence. This study investigated interactions between behavioral and pharmacological strategies in a preclinical model of cocaine vs food choice. Six rhesus monkeys, implanted with a chronic indwelling double-lumen venous catheter, initially responded under a concurrent schedule of food delivery (1-g pellets, fixed-ratio (FR) 100 schedule) and cocaine injections (0-0.1 mg/kg/injection, FR 10 schedule) during continuous 7-day treatment periods with saline or the agonist medication phenmetrazine (0.032-0.1 mg/kg/h). Subsequently, the FR response requirement for cocaine or food was varied (food, FR 100; cocaine, FR 1-100; cocaine, FR 10; food, FR 10-300), and effects of phenmetrazine on cocaine vs food choice were redetermined. Decreases in the cocaine FR or increases in the food FR resulted in leftward shifts in the cocaine choice dose-effect curve, whereas increases in the cocaine FR or decreases in the food FR resulted in rightward shifts in the cocaine choice dose-effect curve. The efficacy of phenmetrazine to decrease cocaine choice varied systematically as a function of the prevailing response requirements, such that phenmetrazine efficacy was greatest when cocaine choice was maintained by relatively low unit cocaine doses. These results suggest that efficacy of pharmacotherapies to modulate cocaine use can be influenced by behavioral contingencies of cocaine availability. Agonist medications may be most effective under contingencies that engender choice of relatively low cocaine doses.