The effects of changes of marital status on changes in health status of elderly respondents over a 14-month interval are examined. Multiple regression is used to examine data on 7696 respondents from the HEW [Department of Health, Education and Welfare, Social Security Administration, USA] 1973-74 Survey of the Low Income Aged and Disabled (SLIAD). Health status is measured by perceived health, days ill in bed at home and days in medical institutions. Significant differences are found in health statuses measured cross-sectionally and longitudinally (net of health status at time 1), among different age, race, sex and educational groups, with better health found among younger respondents, females, whites and the better educated. Marital statuses add significant explained variance (R2) to the longitudinal equations of respondents'' perceived health and days ill in bed, but only to the cross-sectional equation for perceived health. Respondents who lost their spouse between time 1 and time 2 have significantly lower perceived health at time 2 than those who remained married, yet they spend fewer days ill in bed than married respondents. Higher rates of institutionalization are found among those who lost their spouse prior to time 1. Respondents who have never married have better perceived health and spend fewer days ill in bed than married respondents. It is change from married to unmarried status, rathern unmarried status per se, that probably leads to decline in perceived health. Support is also found for the desirability-undesirability explanation of the health consequences of life changes.