Baboons responded on a choice task on which discrete trials involved choosing between an intravenous injection of heroin (.32 or 1.0 mg/kg) or the availability of food pellets. An intertrial interval of three hours followed the completion of each trial. Under baseline conditions baboons consistently completed the eight available trials each day. Typically, animals chose heroin on three or four trials a day and food on the remaining trials. Animals tended to space the selection of heroin rather than choosing heroin on consecutive trials. A series of single-day experimental manipulations was undertaken to characterize performance further. Manipulation of the heroin dose produced shifts in the relative frequency of choosing the drug option which were inversely related to dose. Manipulation of number of pellets per food trial produced little change in distribution of choices. Noncontingent administration of morphine produced dose-related decreases in relative frequency of heroin choices, and a higher dose decreased the number of trials completed. Noncontingent naloxone produced dose-related increases in the relative frequency of heroin choices. Noncontingent secobarbital had no effect on distribution of choices, and high doses reduced the number of trials completed per day. The results suggest that morphine and naloxone produce shifts in this choice behavior by selectively interacting with the reinforcing properties of the option involving heroin.