Three rhesus monkeys lived in primate cages provided with response keys and enclosed in isolation chambers. During experimental sessions on Monday, Wednesday and Friday, the chamber door was closed and every 10th key-pressing response during a 60-minute interval produced a 2-second red light, but had no other programmed consequences (the 10-response fixed-ratio component of the second-order schedule; FR 10). The first FR 10 component completed after the 60-minute interval had elapsed produced a red light which remained on for 2 minutes while the chamber door was opened; the monkey then extended his arm and was given an intramuscular injection of drug (the 60-minute fixed-interval component of the second-order schedule; FI 60 min). Under this second-order schedule of intramuscular drug injection, repeated sequences of rapid responding were maintained during each session by 0.75 to 3.0 mg/kg injections of either morphine or cocaine. Patterns of responding characteristic of FR schedules were controlled by the 2-second red lights; a pause in responding after each 2-second red light was followed by a sustained high rate of responding until the light was produced again. Pauses in responding became progressively shorter as time elapsed in the 60-minute interval. When saline injections were substituted for drug injections, responding markedly decreased. When responding was maintained by 3.0 mg/kg morphine injections, pretreatment with 0.03 mg/kg of nalorphine increased responding, while pretreatment with 0.1 or 0.3 mg/kg of nalorphine decreased responding.