The possible relationship between the incidence of respiratory diseases as reported to general practitioners and numbers of registered deaths in England and Wales has been examined. Morbidity data from sentinel practices for the period 1986-1990 (population covered increased from 220,000 to 470,000) were used to calculate weekly rates of aggregated respiratory disease for persons of all ages and for elderly persons (aged 65 years and over). The elderly respiratory disease rates and numbers of deaths were aggregated into 4-week periods; secular and seasonal trends were removed from each series and the two sets of residuals were examined graphically and cross correlation coefficients calculated. There was a very strong positive association between the respiratory disease rate and number of deaths in the same 4-week period and there was also a significant but less pronounced association between respiratory disease in one 4-week period and deaths in the next. After prior separation of weeks according to temperature into four bands, weekly rates for respiratory disease were also strongly associated with the number of weekly deaths for each temperature band. The synchronisation of peaks and troughs in the two series throughout the year supports the hypothesis that a cause and effect relationship exists between respiratory disease in the elderly and number of deaths. Other climatic and meteorological variables besides temperature may play a part in determining the spread of a respiratory disease. There is a for further research to identify the micro-organisms responsible for acute respiratory infections in the elderly.