BACKGROUND: Wife beating is an important public health problem in many developing countries. We assessed the rates of wife beating and examined factors associated with wife beating in 1995 and 2005 in Egypt. METHODS: We used data from two Demographic and Health Surveys (DHS) conducted in Egypt in 1995 and 2005 using multistage household sampling. Data related to wife beating included information from 7122 women in 1995 and 5612 women in 2005. Logistic regression was used to analyze factors independently associated with wife beating. Special weights were used to obtain nationally representative estimates. RESULTS: In 1995 17.5% of married women in Egypt experienced wife beating in the last 12 months, in 2005--18.9% or 16.0%, using different measures. The association between socio-demographic differentials and wife beating was weaker in the newer survey. The 12-month prevalence of wife beating was lower only when both partners were educated, but the differences across education levels were less pronounced in 2005. Based on the information available in the 2005 survey, more educated women experienced less severe forms of wife beating than less educated women. CONCLUSION: Different measures used in both surveys make a direct comparison difficult. The observed patterns indicate that the changes in prevalence may be masked by two opposite processes occurring in the society: a decrease in (severe forms of) wife beating and an increase in reporting of wife beating. Improving the access to education for women and raising education levels in the whole society may help reducing wife beating.