Diazepam self-administration was studied in rhesus monkeys under several conditions of availability. Lever-press responding was maintained in twelve monkeys under a fixed-ratio 10 (FR 10) schedule of IV cocaine or pentobarbital delivery in daily sessions of 1-3 h duration. Each of several doses of diazepam (0.012-0.4 mg/kg/infusion) or vehicle was periodically substituted for 5-14 consecutive sessions. Between each substitution, responding was maintained by the baseline drug (cocaine or pentobarbital). Another procedure was to decrease the response requirement for drug delivery to a fixed-ratio one (FR 1). In three of eleven monkeys studied under conditions of a cocaine baseline and the FR 10 schedule, responding was maintained by diazepam and was inversely related to dose. In each of five monkeys tested in a similar manner but with a pentobarbital baseline, at least one dose of diazepam maintained responding above vehicle levels. Three of these monkeys had previously failed to self-administer diazepam under the cocaine baseline condition. Subsequently when two of these monkeys were returned to the cocaine baseline, diazepam was not self-administered above vehicle levels. Under FR 1 conditions of substitution, vehicle and pentobarbital intake increased in each monkey tested and cocaine intake increased in two of four monkeys. Diazepam self-administration also increased but did not exceed vehicle levels under the FR 1 schedule. However, in two monkeys the number of diazepam infusions was increased compared to the FR 10 substitution condition. These results emphasize the importance of testing drugs under several conditions to determine their relative dependence potential.