To assess differences in the lag structure pattern between particulate matter <10 &mgr;m/100 &mgr;m in diameter (PM10) and cause-specific mortality, we performed a time-series analysis in 10 US cities using generalized additive Poisson regressions in each city; nonparametric smooth functions were used to control for long time trend, weather, and day of the week. The PM10effect was estimated based on its daily mean, 2-day moving average, and the cumulative 7-day effect by means of an unconstrained distributed lag model. A 10-&mgr;g/m3increase in the 7-day mean of PM10was associated with increases in deaths due to pneumonia (2.7%, 95% confidence interval [CI]: 1.5, 3.9), chronic obstructive pulmonary disease (1.7%, 95% CI: 0.1, 3.3), and all cardiovascular diseases (1.0%, 95% CI: 0.6, 1.4). A 10-&mgr;g/m3increase in the 2-day mean of PM10was associated with a 0.7% (95% CI: 0.3, 1.1) increase in deaths from myocardial infarction. When the distributed lag was assessed, two different patterns could be observed: respiratory deaths were more affected by air pollution levels on the previous days, whereas cardiovascular deaths were more affected by same-day pollution. These results contribute to the overall efforts so far in understanding how exposure to air pollution promotes adverse health effects.