The Behavioral Risk Factor Surveillance System, a data set based on telephone surveys that have been conducted by States in collaboration with the Centers for Disease Control, has been used to estimate the prevalence of behavioral risk factors for adults in the United States so health objectives can be set and progress towards accomplishing them measured. Data for adult American Indians in this regard have not been available generally. The use of these data to estimate behavioral risk prevalence for American Indians by geographic region was examined and the results compared with those for white Americans. In addition, data from the system were compared with other data sets, including the results of selected surveys in American Indian communities, to explore the validity of the system as a tool for evaluating the behavioral risks of Indians. Behavioral Risk Factor Surveillance System data for the period 1985 to 1988 were used. During this period, the 1,055 American Indian respondents constituted 0.63 percent of those responding under the system and 0.70 percent of the population of the participating States. Separate (sex-specific) behavioral risk prevalence estimates were derived for Indians and whites for four geographic regions--Southwest, Plains, West Coast, and Other States. The system's behavioral risk estimates for the Plains region were compared with available data from behavioral risk surveys done in three American Indian communities in Montana (Blackfeet, Fort Peck, and Great Falls) from 1987 to 1989. The behavioral risk factors compared include use of automobile seatbelts, current smoking, current use of smokeless tobacco, heavy drinking, drinking and driving, overweight, hypertension, and sedentary lifestyle. Although large regional differences in the prevalence of these risk factors were found, the magnitude and direction of the differences are frequently similar among American Indians and whites living in the same geographic regions. The findings from the Behavioral Risk Factor Surveillance System among American Indians are largely consistent within dependently collected data from more resource intensive household surveys, at least when surveys in Montana are compared with system data from the Plains. These data are generally consistent with other epidemiologic studies.When they are used in conjunction with community-specific surveys, the Behavioral Risk Factor Surveillance System data may be useful for monitoring the progress of American Indians towards the Year 2000 national health objectives. The value of the surveillance system for monitoring trends in behavioral risk factors among Indians would be enhanced if States attempted to over sample regions (such as Indian reservations) with a high proportion of Indian residents. It appears that aggressive health promotion and disease prevention efforts will be needed if these objectives are to be achieved.