Understanding and prediction of children's adaptation to loss requires attention to family characteristics and interpersonal patterns in addition to individual variables. Empirical inquiries into family variables have indicated that the concepts of cohesion and expressiveness in particular may be useful in explaining members' adjustment. Using both deductive and inductive methods this study developed a reliable behavioral coding system for observing family members as they described the story of a child's death. The study also examined the relationship of these observable behaviors to both self-reports of cohesion and expressiveness and measures of individual adjustment. In terms of convergent validity the Expressiveness subscale of the Family Environment Scale (Moos & Moos, 1986) was more readily associated with observable behaviors than was the Cohesion subscale. Discriminant validity was not established, however, and possible explanations for this were discussed. Examination of criterion-related and predictive validity demonstrated the uti1ity of both se1f-report and behavioral measures of cohesion and expressiveness in accounting for parental depression, child behavior problems, and other specific indicators of distress. Implications for clinical intervention with bereaved families were discussed.