We recently conducted a pilot study supporting the feasibility, safety, and validity of a human laboratory model of ad libitum cocaine administration in which subjects self-selected the timing of infusions. The current study extends this work to include a randomized design with a test-retest component in a larger sample. To investigate the regulation of cocaine intake by humans and its effects on subjective and cardiovascular responses. Subjects were 14 non-treatment seeking volunteers (10 M, 4 F) with cocaine abuse/dependence. Subjects self-administered cocaine infusions (0, 8, 16, and 32 mg/70 kg) over a 2-h period under a fixed ratio 1, 5-min time-out schedule on 4 consecutive days. A fifth session was conducted at 16-mg dose to assess the paradigm's test-retest reliability. Subjects regulated their cocaine intake in a dose-dependent fashion. Self-reports of cocaine-related subjective effects (e.g., "high" and "stimulated") also varied in a dose-dependent way. Test-retest data and the randomized design support the conclusion that such effects are not due to tolerance or other experimental artifacts. The current study replicates prior work demonstrating the feasibility, safety, and validity of our human laboratory paradigm of cocaine administration in a larger sample using a randomized design. The current study also shows the test-retest reliability of these methods, establishing its utility for comparisons of experimental interventions (e.g., pharmacological treatments). Finally, the current study suggests that factors other than drug-induced euphoria (i.e., "high") contribute to the regulation of cocaine-taking behaviors in humans.