OBJECTIVE: To assess the impact of the growth in HMO penetration in different metropolitan areas on the change in the number of generalists, specialists, and total physicians, and on the change in the proportion of physicians who are generalists. DATA SOURCES/STUDY SETTING: The American Medical Association Physician Masterfile, to obtain the number of patient care generalists and specialists in 1987 and in 1997 who were practicing in each of 316 metropolitan areas in the United States. Additional data for each metropolitan area were obtained from a variety of sources, and included HMO penetration in 1986 and 1996. STUDY DESIGN: We estimated multivariate regression models in which the change in the number of physicians between 1987 and 1997 was a function of HMO penetration in 1986, the change in HMO penetration between 1986 and 1996, population characteristics and physician fees in 1986, and the change in population characteristics and fees between 1986 and 1996. Each model was estimated using ordinary least squares (OLS) and two-stage least squares (TSLS). PRINCIPAL FINDINGS: HMO penetration did not affect the number of generalist physicians or hospital-based specialists, but faster HMO growth led to smaller increases in the numbers of medical/surgical specialists and total physicians. Faster HMO growth also led to larger increases in the proportion of physicians who were generalists. Our best estimate is that an increase in HMO penetration of .10 between 1986 and 1996 reduced the rate of increase in medical/surgical specialists by 10.3 percent and reduced the rate of increase in total physicians by 7.2 percent. CONCLUSIONS: The findings of this study support the notion that HMOs reduce the demand for physician services, particularly for specialists' services. The findings also imply that, during the past decade, there has been a redistribution of physicians-especially medical/surgical specialists-from metropolitan areas with high HMO penetration to low-penetration areas.