In July 1993, severe flooding devastated Sarlahi district in Nepal. The next month, a follow-up study of a large population cohort was undertaken. The study is unique in that a prospective research database was used to verify residency prior to the flood and to confirm vital status afterwards. It evaluated 41,501 children aged between two and nine years and adults aged 15-70 in 7,252 households. Flood-related fatality rates were 13.3 per 1,000 for girls and 9.4 per 1,000 for boys, 6.1 per 1,000 for women and 4.1 per 1,000 for men. Flood-related fatality rates for children were six times higher than mortality rates in the same villages a year before the flood (relative risk (RR) = 5.9, 95 per cent confidence interval (CI) 5.0-6.8). Flood-related fatality was associated with low socio-economic status preflood (RR = 6.4, 95 per cent CI 2.7-20.0), and having a house constructed of thatch (RR = 5.1, 95 per cent CI 1.7-24.5).