High-speed chronoamperometry was used to monitor dopamine-related electrochemical signals in the nucleus accumbens of rats allowed to self-administer heroin intravenously and rats that received similar injections passively. Rats self-administered 100 micrograms/kg of heroin at approximately 20-min intervals. Dopamine-related electrochemical signals increased monotonically after the first injection of each day; the effect was weaker on the first than on the second and subsequent days. The second and subsequent injections in each session caused biphasic effects: the initial effect was a decrease in signal--a minor one when compared to the increase caused by the first injection--and this was followed by an increase that brought the signal back to or somewhat higher than the level at the time of the injection. Over the course of each 4-h session, the electrochemical signal reached and fluctuated around an elevated plateau; doubling the injection dose did not elevate this plateau but did cause larger phasic decreases and subsequent increases. Qualitatively similar electrochemical changes were seen in the animals passively receiving the drug, but there were two notable quantitative differences. First, in the passive animals the initial depressions in signal were of shorter duration. Second, in the passive animals (which were injected at intervals determined by the self-administering animals) the electrochemical signal reached a maximum and began to fall prior to the time of the next injection; in the animals that self-administered the drug, the signal was still rising at the time of the next injection. The changes in electrochemical signal are unlikely to represent fluctuations of ascorbate or dopamine metabolites; thus it appears that whereas self-administered heroin injections cause a slow and long-lasting elevation of extracellular dopamine concentration, short-term increases in dopamine concentration are associated with the behavioral activation that precedes the injections and it is short-term decreases that appear to be associated with the period usually thought to be most significant for positive reinforcement.