Daily death counts in Mexico City were examined in relation to ambient ozone levels during 1990-1992 for the purpose of investigating the acute, irreversible effects of air pollution, with emphasis on ozone exposure. Air pollution data were obtained from nine monitoring stations operated by the Departamento del Distrito Federal. Mortality data were provided by the Instituto Nacional de Estadística, Geografía, e Informática. Increases in numbers of deaths were positively associated with elevated air pollution levels on the same day and on the previous day. The magnitude of the increases was small but statistically significant, after Poisson regression models were used to adjust for temperature and long-term trends. In models using data for a single pollutant, the "crude" ratio for total mortality associated with an increase of 100 parts per billion (ppb)* in one-hour maximum ozone concentration was 1.029 (95% CI 1.015, 1.044). A moving average of ozone showed a stronger association (rate ratio [RR] = 1.048, 95% CI 1.025, 1.070), and excess mortality (an increase in the number of deaths, relative to the average on days with low pollution levels) was more evident for persons over 65 years of age. Separate analyses of the effect of elevated ozone for different areas of the city showed similar results, but they were not statistically significant. Other pollutants also were related to mortality. The RR was 1.075 (95% CI 0.984, 1.062) per 100-ppb increase for sulfur dioxide and 1.049 (95% CI 1.030, 1.067) per 100 micrograms/m3 increase in total suspended particulates (TSP) when these pollutants were considered in separate models. However, when all three pollutants were considered simultaneously, only TSP remained associated with mortality, indicating excess mortality of 5% per 100 micrograms/m3 increase [RR = 1.052, 95% CI 1.034, 1.072]. The excess mortality associated with TSP is consistent with that observed in other cities in America and Europe. This study provides some evidence that ozone is associated with all-cause mortality and with mortality among the elderly after controlling for long-term cycles. However, ozone levels exhibited little or no effect on mortality rates when other air pollutants were considered simultaneously. Particulate matter appeared to be an important pollutant; it independently predicted changes in mortality. Nevertheless, because of the complexity and variability of the mixtures to which people are exposed, it is difficult to attribute the observed effects to a single pollutant. The technical feasibility and scientific validity of isolating the effect of single pollutants in such complex mixtures requires further research and careful consideration. Given the large population living in and exposed to ambient air pollution in Mexico City and other metropolises throughout the world, these small but significant associations of mortality with air pollution indices are of public health concern.