In this study, we examined the short-term effects of air mass types on mortality in Athens, Greece. An objective air mass types classification was used, based on meteorological parameters measured at the surface. Mortality data were treated with generalized additive models (GAM) and extending Poisson regression, using a LOESS smoother to control for the confounding effects of seasonal patterns, adjusting also for temperature, long-term trends, day of the week, and ambient particle concentrations. The introduced air mass classification explains the daily variation of mortality to a statistically significant degree. The highest daily mortality was observed on days characterized by southerly flow conditions for both the cold (increase in relative risk for mortality 9%; with a 95% confidence interval: 3-14%), and the warm period (7%; with a 95% confidence interval: 2-13%) of the year. The northeasterly flow is associated with the lowest mortality. Effects on mortality, independent of temperature, are observed mainly for lag 0 during the cold period, but persist longer during the warm period. Not adjusting for temperature and/or ambient particle levels slightly alters the results, which then reflect the known temperature and particle effects, already reported in the literature. In conclusion, we find that air mass types have independent effects on mortality for both the cold and warm season and may be used to predict weather-related adverse health effects.