Parent and child treatment variables posited to bear on reabuse were explicitly defined and evaluated. Reabuse was assessed both on the basis of clinicians' subjective evaluations and official State registry reports of abuse during treatment and following termination. Of 22 variables and their combinations, no single variable by itself was strongly associated with reabuse. However, interactions involving several variables significantly distinguished between reabusers and nonreabusers. These included marital status, income source, official reporting of abuse, and personal history of abuse. For example, while income source did not have a mitigating effect for parents who had themselves been abused, having some earned income decreased the probability of reabuse in those without such a personal history. Similarly, absence of a childhood abuse history among never-married abusers correlated with reabuse. Results suggest that the recurrence of abuse may be more profitably explicated by the investigation of potential joint influences of several variables than by examination of possible factors in isolation from each other. 1 table, 2 figures, and 16 references. (Author abstract modified)