Conceptual and pragmatic issues relevant to the study of persons with autism are addressed within the context of comparison groups and matching strategies. We argue that no choice of comparison group or matching strategy is perfect, but rather needs to be determined by specific research objectives and theoretical questions. Thus, strategies can differ between studies in which the goal is to delineate developmental profiles and those in which the focus is the study of a specific aspect of functioning. We promote the notion of a "mosaic," rather than a "melting pot," approach to science in which researchers communicate conservative and precise interpretations of empirical findings.