The effects of influenza A and B and RSV on mortality in England and Wales were assessed by regression analysis for the period 1975-90. Morbidity data from sentinel practices were used to calculate 4-weekly rates of aggregated upper respiratory tract infections (URTI); PHLS laboratory reports were used as indices of infection, and 4-weekly death rates from all causes, excluding childbirths, were used to study relationships with mortality. Deaths correlated strongly with influenza A and B reports, temperature, and interactions between aggregated URTI and temperature, and RSV outbreaks and temperature. Estimates of 'seasonal' 4-weekly mortality associated with URTI were made by substituting into primary regression models the mean of annual trough consultation rates for aggregated URTI and baseline values for RSV and influenza. Peak 4-weekly mortality associated with URTIs was estimated at c. 24000 and c. 28000 during combined influenza and RSV epidemics of 1975-6 and 1989-90 respectively. Secondary regression analysis was carried out with the estimated 'seasonal' 4-weekly deaths associated with URTI as dependent variable and laboratory data as regressors. Estimated excess mortality associated with influenza was considerable even during years without major epidemics. Overall during the 15 winters the estimated mortality associated with RSV was 60-80% more than that associated with influenza. The modelling permits only a crude estimate of RSV associated mortality. None the less it suggests that RSV is an important cause of winter mortality.