Background Screening strategies for childhood physical abuse (CPA) need to be improved in order to identify those most at risk. This study uses two regionally representative community samples to examine whether a cluster or cumulative model of risk indicators (i.e. parental divorce, parental unemployment, and parental addictions) explains a larger proportion of the variation in CPA. Methods Data were drawn from Statistics Canada's National Population Health Survey (1994–1995) and Canadian Community Health Survey 3.1 (2005). Response rates were greater than 80% in both samples. Each survey had approximately 13 000 respondents aged 18 and over who answered questions about the above adverse childhood experiences. Results A gradient was shown with similar outcomes in each data set. Only 3.4% of adults who experienced none of the three risk indicators reported they had been physically abused during childhood or adolescence. The prevalence of CPA was greater among those who experienced parental divorce alone (8.3%–10.7%), parental unemployment alone (8.9%–9.7%) or parental addictions alone (18.0%–19.5%). When all three risk indicators were present, the prevalence of CPA ranged from 36.0%–41.0% and the age‐sex‐race adjusted odds were greater than 15 times that of individuals with none of the three risk indicators. The cluster model explained a statistically significantly larger proportion of the variation than the cumulative model although the difference between the two models was modest. For the purposes of parsimony, the cumulative model may be the better alternative. Conclusions Adults who were exposed to two or more childhood risk indicators were much more likely to report that they were physically abused during their childhood than those with only one or no risk factors. Medical professionals may use this information on cumulative risk factors to more effectively target screening for potential CPA. Future research should include prospective studies.