Air pollution data from 17 sampling stations between Windsor and Peterborough in Southern Ontario, for January, February, July, and August in 1974 and 1976 to 1983, have been analyzed. Each station reported O3, NO3, SO2, and the coefficient of haze (COH) every hour and aerosol sulfates for a 24-hr period every sixth day using glass-fiber filters. Data on mean daily temperature and relative humidity for the region were also recorded. It is shown that there are high correlations between different pollutants and between these and temperature in the summer. In the summer, sulfate levels were significantly correlated with relative humidity. In winter, the highest correlation was between COH and NO2. Over the 9-year period, SO2 levels in both winter and summer have fallen considerably; there have been no significant trends in O3, NO3, or COH data. Aerosol sulfates increased between 1976 and 1980 in both summer and winter and have since declined slightly. Hospital admission data for the 79 acute care hospitals serving the region, which contains about 5.9 million people, have been analyzed on a daily basis for the same months of the same years. Total admissions and total respiratory admissions have declined about 15% over the period, but asthma admissions appear to have risen. The asthma category of admissions is complicated by the effects of a change in ICD coding in 1979. It has been shown that significant correlations exist between O3, SO4, SO2, and temperature, on the one hand, and deviations from the mean respiratory admissions for that day of the week, for that season, for that year, on the other. These correlations exist if asthma is excluded from the diagnoses. In winter, asthma admissions are correlated with temperature only. A group of nonrespiratory conditions showed no correlations with air pollutants in winter or summer. Stepwise multiple regression analysis based on each year considered individually indicates that in summer SO4 and temperature account for about 5% of the variance in respiratory or asthma admissions. It is shown that the mean of the hourly ozone maxima has a high correlation with the maximal 8-hr average for ozone, and that using this index instead of the mean of the hourly maxima does not increase the correlation coefficient with respiratory disease. Another analysis has been performed by grouping the hospitals and sampling stations into nine separate regions.(ABSTRACT TRUNCATED AT 400 WORDS)