BACKGROUND: There is little information available on nontraumatic health risks as the result of floods, and on the factors that determine vulnerability to them (especially in low-income settings). We estimated the pattern of mortality, diarrhea, and acute respiratory infection following the 2004 floods in rural Bangladesh. METHODS: We conducted controlled interrupted time-series analysis of adverse health outcomes, from 2001 to 2007, in a cohort of 211,000 residents of the Matlab region classified as flooded or nonflooded in 2004. Ratios of mortality, diarrhea, and acute respiratory infection rates in flooded compared with nonflooded areas were calculated by week for mortality and diarrhea, and by month for acute respiratory infection. We controlled for baseline differences as well as normal seasonal patterns in the flooded and nonflooded areas. Variations in flood-related health risks were examined by age, income level, drinking-water source, latrine type, and service area. RESULTS: After fully controlling for pre-flood rate differences and for seasonality, there was no clear evidence of excesses in mortality or diarrhea risk during or after flooding. For acute respiratory infection, we found no evidence of excess risk during the flood itself but a moderate increase in risk during the 6 months after the flood (relative risk = 1.25 [95% confidence interval = 1.06-1.47]) and the subsequent 18 months. CONCLUSIONS: We found little evidence of increased risk of diarrhea or mortality following the floods, but evidence of a moderate elevation in risk of acute respiratory infection during the 2 years after flooding. The discrepancies between our results and the apparent excesses for mortality and diarrhea reported in other situations, using less- controlled estimates, emphasize the importance of stringent confounder control.