Background
The aim of this study was to determine the prevalence of domestic violence by husbands against wives (DVHW) and its correlates using the data of ever married men from a national survey.

Methods
This subset of ever married men (equal to 2823 if weighted and 3165 if unweighted) was extracted from 4297 men (including single) aged 14–54 years, who were interviewed during the Bangladesh Demographic and Health Survey (BDHS) in 2004. The BDHS 2004 applied a pre-tested structured questionnaire for men to collect much information including socio-demographic, life-style, violence (physical and sexual) against wives and reasons for violence. SPSS version 10 was used to perform all of the analyses.

Results
The prevalence of physical, sexual and overall DVHW was 68%, 27% and 72%, respectively within the year preceding the survey. Slapping/twisting (61%) and pushing/shaking/throwing something (40%) were the most common forms of physical DVHW. Age, place of residence, education, religion and number of children showed significant associations (by odds ratio) with at least one form of violence using multiple logistic regression analysis. Both cross-tabs analysis and logistic regression analysis (adjusted for age, place of residence, education, religion and number of children) indicated a significantly higher prevalence and higher odds ratio of DVHW among those husbands who: did not watch television everyday, had experience of pre-marital and post-marital sex, smoked cigarettes/bidi, chewed tobacco and took illicit drugs. The estimates of the lifestyle variables sharply attenuated when all of the variables were included in the same model, although the significance levels were changed for only some of them.

Conclusion
The high prevalence of DVHW deserves immediate attention and intervention in Bangladesh. For effective intervention, the significant correlates of DVHW, especially the lifestyle-related factors, must be taken into consideration to reduce the prevalence and consequences of DVHW in Bangladesh.