The relationship between the number of daily urgent hospital admissions for respiratory illnesses (31 hospitals) and ambient air pollution in Montreal, Canada, was investigated for warm periods between 1984 and 1988. Air pollutants included 1-hr and 8-hr maximum ozone, estimated particulate matter < or = 10 microns in aerometric diameter (PM10), and estimated sulfate (SO4(2-) fraction of PM10. Regression analyses controlled for seasonal and day-of-week trends, autocorrelation, temperature, and relative humidity. For July and August periods, there was a statistically significant (P < 0.01) univariate relationship of all respiratory admissions to 8-hr maximum ozone, which became nonsignificant when coregressed with temperature. Levels of ozone which never exceeded the U.S. National Ambient Air Quality Standard (NAAQS) of 120 ppb and high intercorrelations between ozone, particulates, and temperature may explain this finding. Asthma admissions in the May-October periods increased by 2.7% over mean levels for each 12 micrograms/m3 increase in PM10 levels 3 days prior to the admission day (95% confidence interval, 0.7 to 4.8%). In the July and August periods, hospital admissions for respiratory illnesses excluding asthma were 9.6% higher (95% confidence interval, 0.5 to 18.7%) when estimated SO4(2-) had exceeded 8.1 micrograms/m3 4 days prior to the admission day compared to days when SO4(2-) was at or below this level. There were no significant (P < 0.05) associations of a reference group of nonrespiratory admissions to air pollution after controlling for weather. The above results were found despite levels of PM10 which never exceeded the NAAQS of 150 micrograms/m3. The associations found are relevant to public health, since hospital admissions are expected to be accompanied by considerably more frequent occurrences of less serious outcomes. The present findings suggest that particulate air pollution during photochemically active periods is related to respiratory morbidity in Montreal.