The share of the elderly living with an adult child decreased monotonically throughout the twentieth century, while the probability of reaching old age and the number of years lived in old age increased. As a result, the expected number of life-years lived with adult children while in old age may have increased, decreased, or stayed the same. I estimate that the number of life-years lived in old-age coresidence with adult children stayed roughly constant between 1900 and 1940, while the rate of coresidence declined. Life years lived in old-age coresidence then declined substantially between 1940 and 1990. Moreover, the number of life-years lived in old-age coresidence in 1990 would have been roughly half as great as it actually was had there been no improvements in mortality between 1900 and 1990. And if fertility had remained at its 1900 levels, life-years lived in old-age coresidence would have been about 45% higher in 1990 than it actually was. The results imply that analyses of the change in familial assistance to the elderly should also consider changes in mortality.