BACKGROUND: The importance of respiratory syncytial virus (RSV) infection in adults is not well known, because laboratory testing for RSV infection is not routine. Both RSV infection and influenza are seasonally related, and it is difficult to disentangle one from the other and to disentangle infection from the season and the cold. METHODS: Emergency hospitalizations for respiratory disease from April 1994 to March 2001 were analyzed in relation to surveillance data on RSV infection and influenza, using Poisson regression models adapted for time series and adjusted for season, outdoor temperature, and other covariates. Age-specific admission rates attributable to the viruses were also estimated. RESULTS: Most of the crude relationships of emergency admissions of patients with RSV infection were confounded by season and, to a lesser extent, by cold temperatures. After adjustment for all covariates, including influenza, a 10th-90th percentile increase in RSV counts (defined as the daily number of laboratory reports of RSV) was associated with a rate ratio of 1.36 (95% CI, 1.27-1.45) for emergency admissions for respiratory disease in infants. The rate of hospitalization attributable to RSV infection in children aged <1 year was 5 per 1000 infants per year. The association in people > or =65 years old was much smaller (rate ratio, 1.09; 95% CI, 1.04-1.14; attributable rate of hospitalization , 0.7 per 1000 population); but unlike the association in infants, this association became smaller as the lag (interval) between infection and hospital admission became shorter. Admission rates attributable to influenza were highest in the > or =65-year age group (1.1 per 1000 population), but it was very small at younger ages. CONCLUSIONS: RSV infection appears to be an important determinant of hospitalization in infants in this data set but appears less certain for older persons and requires further investigation.