Intravenous morphine self-injection behaviour was studied using a new discrete-trials procedure in monkeys. Morphine self-injection was most probable when small unit doses (10–100 μg/kg per injection) were available and least frequent, even in comparison with saline substitution, at doses of 300 and 400 μg/kg per injection. Despite variation of the inter-trial interval from 50 to 300 s small injection doses (25 and 100 μg/kg per injection) continued to sustain more frequent self-injection responding than a larger unit dose (400 μg/kg per injection). When daily sessions consisted of 40 trials at an inter-trial interval of 50 s, morphine self-injection behaviour (100 μg/kg per injection) was neither increased nor consistently decreased by naloxone pretreatment (0.1–1.0 mg/kg im). In subsequent experiments, daily sessions consisted of two sets of 40 discrete trials, separated by a time-out (TO) period. When monkeys were pretreated with morphine (5.6 mg/kg im) before a given session, responding for drug injections was suppressed; the effect was reversed by naloxone (1 mg/kg im) which, if injected at TO, reinstated morphine self-administration behaviour. When an injection of chlorpromazine (5.6 mg/kg im) preceded a given session, responding for morphine injections was also suppressed; naloxone (1 mg/kg im) failed to reinstate chlorpromazine-suppressed responding, indicating that naloxone acted as an acute morphine antagonist under some circumstances, but did not otherwise influence behaviour. When the unit dose of morphine was increased to 200 μg/kg per injection, naloxone (1–3 mg/kg im) elicited agitation and occasional vomiting in each monkey tested, but did not suppress their drug-taking behaviour. Though small increases in morphine intake were observed after naloxone pretreatment, substantial increases in response output during the inter-trial interval, with a consequent loss of stimulus control normally exerted by the trial signal, were observed. These results were interpreted in terms of acute, antagonist-precipitated withdrawal signs, not evident in response to naloxone pretreatment when drug self-injection responding was maintained on a daily basis by a limited number of smaller unit injection doses of morphine (50 or 100 μg/kg per injection).