Medical training, practice and research are traditionally organized around body systems and disease categories. There is, however, a disciplinary split over the question of whether the clinical diagnosis is the central issue in describing an individual with an illness. Data from two studies, one institutional and one population based (The Pediatric Ambulatory Care Treatment Study and the National Health Examination Survey--Cycles II and III), are used to test the usefulness of diagnostic groupings in examining correlates of illness. A series of analyses of variance with the diagnostic groupings as the independent variable and a range of psychological, social and educational measures as the dependent variables reveal only the number of significant differences expected by chance. The only area in which a pattern of significant differences is found in the family's interaction with the health care delivery system. These results indicate that there is more variability within diagnostic groupings than between them and suggest that diagnosis is not a helpful categorization in the examination of psychological and social variables. While not surprising to social scientists, these data suggest the need for a major reorientation of the research paradigm when examining the psychological, social, rehabilitative and preventive issues raised by chronic illness in children and families.