To analyze the age-related enrollment of cancer patients onto registration trials of new drugs or new indications approved by the US Food and Drug Administration from 1995 to 2002. This study involved retrospective analyses of demographic data of cancer patients enrolled onto registration trials. The data on 28,766 cancer patients from 55 registration trials were analyzed according to age distributions of > or = 65, > or = 70, and > or = 75 years. The rates of enrollment in each age group for each cancer were compared with the corresponding rates in the US cancer population. The age distributions of the US cancer population were derived from the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute for the period 1995 to 1999 based on the 2000 US Census. The proportions of the overall patient populations aged > or = 65, > or = 70, and > or = 75 years were 36%, 20%, and 9% compared with 60%, 46%, and 31%, respectively, in the US cancer population. Statistically significant under-representation of the elderly (P or = 70 years accounted for most of the under-representation. Elderly were under-represented in the registration trials of new cancer therapies. Various strategies may be needed to evaluate cancer therapies for the elderly in prospective clinical trials and to improve cancer care in the elderly population.