A Poisson specification of the relationship between atmospheric pollution and lost work days is estimated. An important feature of the procedure is control for city-specific effects. A major source of ambiguity in interpreting the results of observational data on pollution versus health status or death rates is that pollution in a city may be correlated with other characteristics ofthat city that affect these outcomes but are not controlled for in the analysis. Or, individual attributes of residents may be correlated with pollution levels but notaccounted for in the analysis. Our results suggest a statistically significantand quantitatively important effect of total suspended particulates on work days lost. A standard deviation increase in total suspended particulates is associated with approximately a ten percent increase in work days lost. As a concomitant of our analysis, we also find a substantial relationship between smoking by others in the individual's household and work days lost by non-smokers.