BACKGROUNDCancer prevalence—the proportion of a population with cancer, including those recently diagnosed, those in treatment, and survivors—is an important indicator of future health care requirements. Only limited information on cancer prevalence is available for the United States. In particular, comparative interstate studies are not available. In this study, we estimate and analyze the prevalence of seven major cancers in Connecticut, lowa, and Utah using the tried and tested PREVAL method applied to National Cancer Institute registry data.METHODSWe analyzed data on 242,851 carcinomas of the stomach, colorectum, pancreas, breast, uterus (corpus), ovary, and non‐Hodgkin lymphoma (NHL), diagnosed in white Americans from 1973 through 1992. Observed prevalence was estimated by applying the PREVAL method to incidence and life status data from the cancer registries. Complete prevalence was estimated by applying correction factors obtained by modeling incidence and survival rates.RESULTSThe ratio of the highest to the lowest prevalence (as proportions) ranged from 1.69 for uterine carcinoma to 2.73 for stomach carcinoma, showing that marked differences in cancer prevalence exist within the United States. Utah had the lowest prevalence for each carcinoma. Connecticut and lowa had similar prevalence levels for carcinomas of the colorectum, pancreas, and ovary and for NHL. Breast carcinoma was the most prevalent, with 826 cases per 100,000 of population in Utah, 1518 per 100,000 in lowa, and 1619 per 100,000 in Connecticut. Cancer survival did not differ greatly among the three registry populations. The major determinants of prevalence differences were incidence and the population age distribution.CONCLUSIONSPREVAL provides reliable estimates of the numbers of living people in a population who have had a cancer diagnosis. Prevalence depends on incidence and survival and on the age structure of population. All these factors have changed markedly in recent years and will continue to do so in the future. Cancer prevalence should be monitored over time to evaluate changes by area, sex, age, and cancer site. The prevalence figures presented are directly comparable with those from European cancer registries. Cancer 2002;95:430–9. © 2002 American Cancer Society.DOI 10.1002/cncr.10675