I used generalized additive models to analyze the time series of daily admissions for chronic obstructive pulmonary disease (COPD) over the period 1987-1995 in three major metropolitan areas, Cook County, Los Angeles County, and Maricopa County, in the United States. In Cook and Maricopa counties I had admissions information only for the elderly (ages 65 yr and over). In Los Angeles County I had admissions information for all ages. In all three counties I had monitoring information on PM10, CO, SO2, NO2, and O3. In Los Angeles County, I had information on PM2.5 in addition. In Cook and Maricopa counties there was weak evidence of an association between the gaseous pollutants other than ozone and admissions. There was no evidence of an association with PM. In Cook and Los Angeles counties, in single-pollutant models, ozone was associated with admissions during the period April-September but not in full-year analyses. In Los Angeles, the other gases were strongly associated with COPD admissions. PM was also associated with admissions in single-pollutant models. The coefficients for PM were greatly attenuated and became insignificant in joint analyses with any one of the gases (with the exception of ozone). Analyses in three broad age ranges (0-19, 20-64, and 65 yr and over) yielded similar results. The results indicate that the gases, other than ozone, were more strongly associated with COPD admissions than PM and that there was considerable heterogeneity in the effects of individual pollutants in different geographic areas of the country.