In an attempt to evaluate periodic health examinations, 10 clinics supplied data on people known to have died while in such a program. The cause of death was ascertained and pertinent data obtained from the medical records of these people. The proportion of deaths in which the cause of death was diagnosed ante mortem was calculated by cause of death, by various other characteristics, and for a matched sample of people still living. In only 51% of the 350 deaths was the cause of death recognized on or before the last examination. This percentage recognition was highest for heart disease other than coronary and lowest for malignant disease. It was higher in the old than in the young, greater when last examination was performed closer to death than when the interval was longer, and was much higher in non-smokers than in heavy cigarette smokers.