Members of a group of hospitalized children under 4 years of age (48 under care for child abuse, 97 for accidents, 41 for failure to thrive, and 23 for ingestions) were matched individually on age, social class, and race with controls suffering from comparable acute medical conditions. A structured maternal and paternal interview yielded 63 significant single-variable comparisons in these domains: child health and development, past and present family disruption and conflict, parental physical and emotional health, and environmental setting. Discriminant function analysis suggested interrelationships among the case groups and an additive mode of pathogenesis, with more severe stresses associated with more severe childhood symptoms. A classification analysis was performed to determine the extent to which equations generated in the discriminant function analyses successfully differentiated between patients in the illness categories and in the control groups. Cluster analysis on a random half-sample identified three cohesive groups, characterized as "ecologic equilibrium,""adversity," and "crisis." This reformulation subsumed respectively increasing proportions of severe symptoms and replicated successfully on the other half-sample. Its elaboration gives a convenient matrix for organizing data from practice and a value-free alternative to the present manifestational classification system. Additionally, a statistical analysis of group profiles was performed, and high-risk families were profiled in nine vital aspects of their lives. In nearly every aspect the crisis group was found to suffer. (RH)