This study examined the effects of biosocial variables on changes in nutritional status of rural Bangladeshi children, aged less than 2 years, pre- and post-1987 monsoon flooding. Nutritional status was measured by weight for age: variables included were age, sex, sickness during 2 weeks preceding the survey, intake of vitamin A capsules, socioeconomic status of household, and mother's education. A multivariate logistic regression analysis revealed an adverse effect of flood on nutrition and the effect was dependent on sex of child and intake of vitamin A. After the flood the proportion of severely malnourished children was significantly greater among those who had not taken vitamin A. For boys the proportion with severe malnutrition increased after the flood and the increase was greater than for girls; however, boys always had a lower risk of severe malnutrition than girls.