Insurers and employers perceive the demand for mental health care to be highly responsive to the terms of insurance. Better coverage, it is believed, would increase demand, increasing expenditures through use of services that may be discretionary in nature. This article attempts to shed light on this issue by summarizing and evaluating the results of more than 40 published and unpublished studies. The major criterion for inclusion was the availability of information on the size of the population covered, so that rates of utilization could be calculated. More recent studies are emphasized. If research at the population level using aggregate utilization as a dependent variable is the "first generation of research," studies of individual use over a period of a year constitutes the "second generation." The emerging research on episodes of treatment represents a new "third generation" of studies. If some progress can be made on issues of ways in which patients form expectations about their treatment and its cost, this new generation of research promises to model demand response more precisely to coverage terms that change within a year, such as deductibles or limits.