OBJECTIVE: To analyze primary care staffing in HMOs and to review the literature on primary care organization and performance in managed care organizations, with an emphasis on the delivery of primary care to the elderly and chronically ill. DATA SOURCES/STUDY SETTING: Analysis of primary care staffing: InterStudy HMO census data on primary care (n = 1,956) and specialist (n = 1,777) physician staffing levels from 1991 through 1995. Primary care organization and performance for the chronically ill and elderly were analyzed using a review of published research. STUDY DESIGN: For the staffing-level models, the number of primary care and specialist physicians per 100,000 enrollees was regressed on HMO characteristics (HMO type [group, staff, network, mixed], HMO enrollment, federal qualification, profit status, national affiliation) and community characteristics (per capita income, population density, service area size, HMO competition). For the review of organization and performance, literature published was summarized in a tabular format. PRINCIPAL FINDINGS: The analysis of physician staffing shows that group and staff HMOs have fewer primary care and specialist physicians per 100,000 enrollees than do network and mixed HMOs, which have fewer than IPAs. Larger HMOs use fewer physicians per 100,000 enrollees than smaller HMOs. Federally qualified HMOs have fewer primary care and specialist physicians per 100,000 enrollees. For-profit, nationally affiliated, and Blue Cross HMOs have more primary care and specialist physicians than do local HMOs. HMOs in areas with high per capita income have more PCPs per 100,000 and a greater proportion of PCPs in the panel. HMO penetration decreases the use of specialists, but the number of HMOs increases the use of primary care and specialist physicians in highly competitive markets. Under very competitive conditions, HMOs appear to compete by increasing access to both PCPs and specialists, with a greater emphasis on access to specialists. The review of research on HMO performance suggests that access to PCPs is better in MCOs. But access to specialists and hospitals is lower and more difficult in MCOs than FFS. Data do not suggest that processes of care, given access, are different in MCOs and FFS. MCO enrollees are more satisfied with financial aspects of a health plan and less satisfied with other aspects of health plan organization. There are potential problems with outcomes, with some studies finding greater declines among the chronically ill in MCOs than FFS. We found a variety of innovative care programs for the elderly, based on two fundamentally different approaches: organization around primary care or organizing around specialty care. Differences between the performance of the two approaches cannot be evaluated because of the small amount of research done. It is difficult to say how well particular programs perform and if they can be replicated. The innovative programs described in the literature tend to be benchmark programs developed by HMOs with a strong positive reputation.