Four bases, with variants, of estimating the number of centenarians in the United States are described: (a) direct use of Medicare records; (b) the forward-survival method; (c) the vital statistics method; and (d) population reconstruction using death statistics. Estimates for each year, 1950, 1960, and 1970, for sex and race groups are presented separately and compared with the census counts. The plausibility and accuracy of the estimates derived by the various methods are discussed, as well as possible reasons for the overstatement of centenarians in recent censuses.