Many recent analyses have reported associations between air pollution and mortality in U.S. cities. In this paper, we present the results of regression analyses of daily mortality in Philadelphia during the period 1973-1988. Pollution variables included in the analyses were total suspended particulates (TSP), sulfur dioxide, and ozone. We controlled for the effects of weather on mortality by analyzing mortality separately for each season and explicityly including quintiles of temperature in the regression models. In regression models that consider weather and pollution variables simultaneously, daily mortality is associated with hot days in summer [relative risk (RR) for highest quintile of temperature = 1.07; 95% confidence interval (CI) = 1.04-1.10], and with cold days in spring (RR for lowest quintile of temperature = 1.07; 95% CI = 1.04-1.10), fall (RR for lowest quintile of temperature = 1.05; 95% CI = 1.02-1.08), and winter (RR for lowest quintile of temperature = 1.04; 95% CI = 1.01-1.07). When all three pollution covariates and weather are considered simultaneously in the regression model, ozone is associated with mortality in summer (RR = 1.15; 95% CI = 1.07-1.24) and sulfur dioxide is associated with mortality in spring (RR = 1.19; 95% CI = 1.06-1.33), fall (RR = 1.14; 95% CI = 1.00-1.29), and winter (RR = 1.21; 95% CI = 1.09-1.35), where the relative risks are for incremental changes of 100 parts per billion in ozone and sulfur dioxide on the previous day.(ABSTRACT TRUNCATED AT 250 WORDS)