BACKGROUND: This paper compares husband and wife reports of wife beating using household survey data collected from poor Palestinian refugee communities in Lebanon. METHODS: The analyses are based on a matched data file of 417 currently married couples, drawn from a unique multi-purpose living conditions sample survey of about 3600 Palestinian refugee households interviewed in the spring and summer of 1999. Four outcomes (ever beaten, last year beating, beating during pregnancy, and injuries caused by beating) were analysed using Kappa statistics and per cent agreement. Logistic regression was used to analyse discordant reporting of wife beating during the year preceding the survey. RESULTS: Husband and wives' reports of the four different outcomes are in 'good' agreement as judged by Kappa coefficients, ranging from 0.62 for 'beaten during pregnancy' to 0.69 for 'injuries resulting from beating'. Prevalence estimates of domestic violence are also remarkably similar. However, findings from a multivariate logistic regression model on agreement regarding 'last year beating' show that only age of men was a significant predictor of agreement, controlling for education level, marital duration, region of residence, household size, health status, and consanguinity. CONCLUSIONS: Our findings show that men's self-reports of their violent behaviour against their wives are fairly congruent with those of their spouses, implying that the perpetrators, men, can be 'trusted' in providing basic information on 'beating histories' in epidemiological and demographic population-based investigations in contexts similar to ours. However, care should be taken in studies of young men's current beating behaviour using only their self-reports.