As researchers have examined the so-called "behavioral phenotypes" of different genetic mental retardation conditions, the measurement of behavior becomes increasingly important. From earlier, more global examinations of noteworthy behaviors or profiles, recent researchers have moved from reliance on a single, all-encompassing questionnaire to more multimethod, multireporter approaches. This paper summarizes strengths and weaknesses in utilizing parent-report questionnaires and checklists; psychiatric nosology and specific psychiatric questionnaires; observational studies; reports from individuals with mental retardation themselves; profile analyses of psychometric test and subtest scores; and "grouping" analyses of items (factor analysis) and people (cluster analysis). Although we advocate a mixture of approaches to behavioral assessment, in future work researchers will require more theoretically driven methods for studying behavior, greater acknowledgment that the child's behavior is simply one event in ongoing interactions-transactions with other people, and more attention to age-related changes in behavioral phenotypes.