The precise number of deaths due to rotavirus infection in England and Wales is difficult to determine because mortality is usually low and reported simply as death due to gastrointestinal disease. We have used two independent methods to estimate this figure. Firstly, we obtained records from the Office for National Statistics of deaths in children with any mention of gastroenteritis. For deaths with uncertain aetiology, we estimated the proportion attributable to rotavirus using a linear model with monthly laboratory reports of gastrointestinal micro-organisms as predictors. Secondly, we calculated the case-fatality ratio of records in Hospital Episode Statistics with rotavirus indicated, and applied this ratio to an estimate of the actual number of rotavirus hospital admissions. The two methods gave estimates of 3.8 and 3.2 rotavirus deaths a year respectively.