Historic air pollution episodes of the 1950s led to acute increases in infant mortality, and some recent epidemiologic studies suggest that infant or child mortality may still result from air pollution at current levels. To investigate the evidence for such an association, we conducted a time-series study of infant mortality in the southwestern part of Mexico City in the years 1993 to 1995 using mortality data from death registrations and air pollution measurements from a monitoring station we operated. Excess infant mortality was associated with the level of fine particles in the days before death, with the strongest association observed for the average concentration of fine particles during the period 3 to 5 days previously: a 10-microg m(-3) increase in the mean level of fine particles during these 3 days was associated with a 6.9% excess of infant deaths (95% confidence interval 2.5-11.3%). Infant mortality was also associated with the levels of nitrogen dioxide and ozone 3 to 5 days before death, but not as consistently as with particles.