Unexplained autumn increases in hospital admissions for asthma have been reported in many countries, including the United States, Canada, England and Wales. To investigate the role of infection, the association was tested between hospital admissions for asthma and respiratory infections among preschool children in Metropolitan Toronto, Canada during the period 1981 to 1989. The seasonal pattern in overall hospital utilization was assessed by admissions for nonrespiratory diseases. Time series analysis was used to remove potentially confounding temporal trends and the influence of correlated errors. A fourfold increase in asthma admissions occurred between July and October unaccompanied by similar increases in nonrespiratory admissions. Admissions began increasing during the third week of August, peaked during the third week in September, and slowly decreased during November and December. After adjusting for serial correlation, trends, climate, ambient air pollution and aeroallergens, the seasonal pattern of respiratory infection explained 14% of the variance in asthma admissions. Based on seasonal patterns, respiratory infection is the major identifiable risk factor for the large autumnal increase in asthma admissions.