Second-order schedules of drug self-administration were developed to incorporate the effects of drug-related environmental stimuli into an animal model of drug abuse, making it more similar to human situations. Ironically, little is known about how human subjects behave under these schedules. In this study, human volunteers with a history of cocaine use worked on a second-order schedule in which every 100th lever response produced an auditory-visual brief stimulus (2 s). The first stimulus produced after 1 h was extended to 10 s and paired with an intravenous injection of cocaine (25 mg). Up to three injections were allowed per session. In different phases of the experiment, presentation of the brief stimulus was discontinued and/or saline solution (placebo) was injected instead of cocaine. Injections of cocaine were found to maintain responding even when the brief stimulus was not presented. Placebo injections alone did not maintain responding. In contrast, the brief stimulus maintained high levels of responding under placebo conditions, even though self-reports indicated that subjects could clearly discriminate that they were not receiving cocaine. These results demonstrate that drug-related environmental stimuli can maintain persistent drug seeking during periods of drug unavailability. As this procedure directly measures the effects of stimuli on drug seeking, it may provide a valuable complement to indirect measures, such as self-reports of craving, that are often used with human subjects. The similarity of the response patterns in humans and animals also supports the use of second-order schedules in animals as a valid model of human drug seeking.