To determine whether death can be postponed until after an important social occasion the number of deaths before and after the Jewish holiday of Passover (1966-84) were compared. Passover was chosen for study because it allows comparison of the participating and non-participating (control) groups, and it moves around the calendar (thus allowing separation of the effects of the holiday from the effects of the seasons). In the total Jewish sample (n = 1919), the number of deaths was lower than expected in the week before Passover and higher than expected in the week after (p = 0.045). This dip-peak pattern of mortality was concentrated among people with unambiguously Jewish surnames (p = 0.003) and did not appear in various control groups, including Blacks, Orientals, and Jewish infants. The pattern was most pronounced in the years when the holiday fell on a weekend, when it is most likely to be celebrated by the largest number of people (n = 183; p = 0.001). This pattern was statistically significant by binomial tests and by regression analyses of the data. The Passover pattern of mortality was found in each of the three leading causes of death.