This chapter reviewed the controlled tricyclic trials for cocaine abuse with both a clinician's and a researcher's eye in order to develop methodologic recommendations for future medications development efforts. The review is summarized in table 2. The main points are that attrition is high, particularly early in the trials; placebo effects are high, particularly early and in subjective or self-report measures; and the samples may be heterogeneous with responsive (depressed, mild severity) and unresponsive (antisocial personality) subgroups. Methodologic recommendations are summarized in table 3. Emphasis is placed upon the potential heterogeneity of cocaine abusers and targeting treatments to subgroups on the one hand, and various methodologic recommendations to tighten up the design of early, small-scale pilot trials on the other. These include use of potent, standardized interventions to reduce attrition; a prolonged, single-blind placebo lead-in to wash out early dropouts and placebo effects; discarding the uncontrolled pilot trial in favor of crossover, discontinuation, or multiple-baselines designs; and considering the impressions of experienced clinicians as well as objective, urine-based measures when judging efficacy. These recommendations are all arguable in that they have disadvantages as well as advantages and that they all depart to some extent from current practice and wisdom. It is hoped that they will promote discussion and stimulate methodologic innovation in the search for effective medications for cocaine abuse.