Prologue: During congressional hearings to validate the precision of estimates of the impact of the Clinton health reform plan on the federal budget, Congressional Budget Office (CBO) Director Robert Reischauer patiently fielded dozens of questions. One member of Congress wished to know if the CBO's estimates were “in the ballpark.” Reischauer replied, “Congressman, I believe that we are in the town the ballpark is in.” The flippancy of this comment does not belie its accuracy. Indeed, analysts charged with determining the fiscal impact of the Clinton plan found themselves faced with an enormous task and impossibly short deadlines. They also possessed an awesome responsibility, because legislation could rise or fall in Congress based on the CBO's “scoring” of a particular health plan. In this paper Reischauer and his colleague Linda Bilheimer discuss the perilous path of cost estimation, addressing what estimators need to know and whether the data exist in rich enough detail to provide an adequate estimate on which to base policies that will affect the lives of millions of Americans. In their view, the available data are woefully inadequate to produce good estimates of such complex factors as the impact of the health insurance plan choices of diverse populations, the extent to which risk segmentation affects costs, and other issues for which a simple solution does not exist. The comments of three veteran estimators—Len Nichols, John Sheils, and Ken Thorpe—which follow Bilheimer and Reischauer's paper, underscore this point. Bilheimer is deputy assistant director of the CBO; she holds a doctorate in economics from Harvard. Reischauer was director of the CBO from March 1989 until March 1995. He holds a doctorate in economics from Columbia University and is now a senior fellow at The Brookings Institution.