Parity is associated with mortality among middle-aged women, while substantially less is known about this relationship for men and the elderly. Using the census-based Israel Longitudinal Mortality Study (ILMS) II (1995-2004) we sought to examine the parity-mortality relationship among men and women, middle-aged and elderly. In our study cohort of 71,733 married men and 62,822 married women ages 45-89 years at baseline, 19,437 deaths were reported. Mortality differentials by parity were assessed using Cox proportional hazard regression models adjusted stepwise for age, origin, education and number of rooms. Analyzes were carried out for middle-aged (45-64 years) and elderly (65-89 years) men and women separately. We observed a non-linear relationship between parity and mortality for all individuals even after adjustment for demographic and socio-economic variables. In fully adjusted models, for example, nulliparous middle-aged women experienced the highest mortality risks (hazard ratios [HR] = 1.57, 95% confidence intervals [CI] 1.24, 1.98) followed by those with one child (HR = 1.29, 95% CI 1.10, 1.51). These results were attenuated somewhat for nulliparous older women (HR = 1.25, 95% CI 1.11, 1.41). The detrimental effects of low and high parity on mortality among both men and women suggest a non pregnancy-related pathway that is likely mediated by biological and psychosocial factors and other lifestyle characteristics that have long-term consequences into older ages. Further research is warranted to examine the effects of parity by specific cause of death.