The two examples provided in this chapter suggest that the inconsistent findings across studies evaluating identical pharmacologic agents may be associated with variations in sample characteristics, particularly those associated with (a) general treatment responsiveness (e.g., severity of cocaine use, sociopathy), or (b) responsiveness to specific treatment strategies (e.g., rates of depression where antidepressant agents are evaluated). Describing study samples and evaluating treatment response along multiple dimensions, using a common set of standardized assessments, would be an important advance in understanding variation in subjects' response to medication effects and comparison of findings across different studies. Moreover, consistent description of study samples across a number of dimensions would set the stage for meta-analyses of patient-treatment interactions. Similarly, as new medications are developed and evaluated, variables that have a theoretical basis as mediators of treatment response should be identified and evaluated. It should be noted, however, that success profiling and matching research is more complex than the search for simple main effects (Finney and Moos 1986; Project MATCH Research Group 1993). In particular, adequate power to detect patient-treatment interactions requires much larger sample sizes than those that have to date characterized pharmacotherapy research for cocaine dependence. This strategy, however, is likely to enhance the development of effective pharmacological interventions for this very challenging patient population as researchers' understand the complex processes associated with treatment seeking, retention, and outcome among cocaine abusers.