First, the relationship between temperature and mortality risk from cardio-vascular, respiratory, and other diseases was estimated for each city; then the variation in risk across cities, due to difference in latitude and other variables, was examined.