Earlier studies of cocaine's effects on the hypothalamic-pituitary-adrenal (HPA) axis used nonresponse-contingent designs in which the investigator determined dose, timing, and route of administration. It is important to evaluate whether "control" over cocaine delivery is a significant determinant of cocaine's HPA axis effect. This study measured cocaine's effects on plasma adrenocorticotropic hormone and cortisol, using nonresponse-contingent injections followed later by response-contingent cocaine delivery. In addition, the effects of cocaine history on the HPA response to a noncontingent injection of 1 mg/kg of cocaine were measured. HPA effects of corticotropin-releasing hormone (CRF) were also measured. Male and female rhesus monkeys, with surgically placed venous catheters, were tested in their home cages. Up to 13 injections of saline and cocaine (0.01-, 0.03-, 0.1-, and 0.3-mg/kg/injection) were administered at 10-min intervals (nonresponse-contingent condition) and on a fixed ratio 30, time out 10-min schedule of reinforcement. Overall, cocaine delivered response contingently produced larger, more dose-dependent HPA responses than did noncontingent delivery. The HPA response to a 1 mg/kg cocaine infusion in cocaine-naive monkeys was not predictive of the HPA effect of this dose subsequent to acquisition of cocaine self-administration. Overall, male monkeys had larger HPA responses to cocaine than did female monkeys. Finally, the HPA effects of CRF were significantly correlated with those of large cocaine doses delivered nonresponse contingently, but not with response-contingent administration.