The author offers insights into how the proliferation of competitive health care financing and service delivery systems based on managed care affects the financial support available to academic medical centers (AMCs), especially to their programs in graduate medical education (GME). The paper is based largely on case studies of AMCs conducted by the author in the summer of 1994 in the health care markets of San Diego, California, Minneapolis-St. Paul, Minnesota, and Washington, D.C., complemented by a review of the literature. In sum, the investigator found consensus among all parties that in the current market, managed care plans neither are willing nor feel able to pay much, if any, premium for the services of AMCs, particularly when established, respected alternatives exist, as they typically do for most services in major urban markets. Relatively few short-term adverse effects on AMCs were found from the growth of competitive systems, but AMCs are nevertheless very concerned that managed care will put them at a disadvantage. They are thus seeking ways to position themselves for the future. The AMCs are concerned that at some point, the cost reductions they are making will hinder the fulfillment of their unique traditional mission, since they believe that the costs of their GME programs can be reduced only so far without harming residents' training. Many managed care plans, however, question the AMC mission, taking issue particularly with the training AMCs provide and its relevance to current needs for primary and ambulatory care. The investigators also found considerable support for pooled funding for GME among diverse parties, but no consensus on how this funding should be structured, who should receive it, or what it should support. Potential conflicts were also identified between national, state, and market objectives for provider supply and specialty distribution because these objectives can embody different criteria for assessing the handling and locations of specialists' training. In addition, the findings indicate that it could be unwise to consider AMC policy independent of workforce objectives; doing so could create conflicts about the kinds of physicians who should be trained. The author concludes with a list of approaches to future research that may be constructive.