In this study a rat self-administration model was used to examine the effects of training dose and time in the session on the dose-effect curve for heroin. Doses of heroin lower than 5.4 microg/inf maintained higher rates of drug intake in animals trained with 5.4 microg/inf compared to 18 microg/inf. Doses greater than 5.4 microg/inf maintained similar rates of intake in both groups of animals. The dose-response curve was shifted downward and to the right as the session progressed for animals trained with 5.4 microg/inf of heroin; however, the shift in the dose-intake curve over the session was less pronounced when the training dose was 18 microg/inf. Naltrexone and naltrindole were administered to animals in which responding was engendered with infusions of 5.4 microg of heroin to determine the effects of these antagonists in the context of time is the session. The potency of naltrexone decreased across the 4 h of the session with a time course that was consistent with literature reports on the elimination kinetics of naltrexone in rat brain. In contrast, there was not a significant interaction between naltrindole dose and session time. Therefore, the rates of heroin intake in rats are dependent not only upon the dose available for self-administration, but upon the session time and training dose as well.