A longitudinal design is proposed to investigate the short-term effects of air pollution on health status. The study design exploits a new statistical methodology developed by Dewanji and Moolgavkar (2000, 2002) that makes use of a Poisson counting process for the incidence of events in a longitudinal cohort study. The methodology allows for modeling of subject-specific baseline hazards of multiple events and time-dependent ecological (i.e., daily levels of air pollution and weather conditions) and individual covariates (e.g., comorbid conditions, indices of declines in health status). The study will determine the association between short-term changes in air pollution and incidence of hospitalization, emergency department visits, "intermediate health conditions" (e.g., changes in health status reflected by changes in filled prescriptions), and mortality. Data from the universal Quebec medicare system are used, including data for residents of Montreal age 65 yr and above. The cohorts include individuals who have at baseline airways disease, chronic coronary artery disease, congestive heart failure, hypertension, cerebrovascular disease, atherosclerosis, dysrhythmias, and diabetes, and these will be followed for the incidence of the already mentioned adverse outcomes, adjusting for time-dependent individual and ecological covariates. This type of study will provide findings that can be used to confirm or refute results of time series analyses. Using a different methodology that includes subject-specific information will greatly increase our understanding as to whether short-term exposures to ambient air pollution cause serious changes in health status among subgroups of the population. These results will also help in understanding mechanisms by identifying groups that are susceptible to the effects of air pollution.