Background: Previous studies have found that multiple factors are associated with parent-child relationship quality, but have not distinguished potential sources of between-family and within-family variation in parent-child relationship quality. Methods: Approximately equal numbers of biological (non-stepfamilies), single-mother, stepfather, and complex/stepmother stepfamilies were selected from a large community study in England. The sample comprised 404 children in 171 families. Parent-child warmth/support and conflict/negativity were assessed using questionnaire and interview methods; family socio-demographic data were collected from parent report. Results: Multilevel model analyses indicated substantial within-family variation in parent-child relationship quality, and greater within-family variation in complex/stepmother families compared to other family types. Within-family variation was largely accounted for by differences in the siblings' biological relatedness to the mother and father and to child-specific factors, notably aggressive behavior. Conclusions: The findings illustrate how research using multilevel model designs and analytic strategies may enhance our understanding of family process.