Long-term survival of 3,173 patients with cancer of the uterine cervix and 2,394 patients with cancer of the uterine corpus is compared with survival in the general population to get relative rates. Ratios of observed-to-expected survival rates, called relative survival rates, are given for periods up to 20 years after diagnosis and are analyzed with respect to age at diagnosis, stage of disease, treatment, and other variables. For each site, relative survival rates were low at first but rose to about 97 percent per year after 5 years and showed no further improvement. The effects of age and stage on relative survival rates were small after the first few years. After the 5th year, relative survival of patients with cervical tumors was the same whether the original treatment was radiation, surgery, or radiation plus surgery. The group of corpus-cancer patients treated with surgery and the group treated with radiation plus surgery had survival rates only moderately below the expected values. However, for the group treated with radiation alone, survival rates were significantly below expected levels for at least 20 years. In all treatment groups and for both sites, excess mortality was partly due to late recurrences of uterine cancer. The continuing high mortality of women with corpus tumors treated only by radiation appears to be partly due to the selection of more favorable patients for surgery, but the data suggest the existence of other factors. Whatever the explanation may be, patients with endometrial cancers treated solely with radiation are subject to a substantially greater risk of late recurrence than patients treated surgically.