A non-concurrent prospective study in Washington County, Maryland identified 4,032 (1,204 male, 2,828 female) White persons aged 18 and over who were enumerated in a 1963 non-official census and who became widowed between 1963 and 1974, and an equal number of married persons, each matched to a widowed as to race, sex, year of birth and geography of residence. All were followed to 1975, the date of a second census. Mortality rates based on person-years at risk were about the same for widowed as for married females, but significantly higher for male widowed than male married, even after adjustment for a number of demographic, socioeconomic, and behavioral variables. Mortality rates among widowed males who remarried were very much lower than among those who did not remarry, but no significant difference was observable among widowed females who did nor did not remarry. Multiple regression analysis also showed that, for both sexes and independently of other factors, moving into a nursing home or other chronic care facility was associated with higher mortality than any other residential change or no change, and living alone was associated with higher mortality than living with someone else in the household.