Abstract : Notwithstanding determined efforts during the 1960s to improve the outcome of major system acquisition programs by altering contractual approaches and by introducing a variety of management reforms, typical programs continued to exhibit an average cost growth of about 40 percent (corrected for quantity changes and inflation), a schedule slip of about 15 percent, and final system performance that was likely to deviate by 30 or 40 percent from the original specification. Such findings have been reported in earlier studies. During the present study, an examination of the basic factors contributing to cost growth suggested two avenues for improvement: 1. The cost estimation process could be improved so that cost predictions made early in a program would more nearly correspond to the cost actually experienced. 2. More fundamental improvements seem likely to be achieved through some basic changes in the acquisition process. A potentially important step in cost estimation improvement would be to incorporate explicitly in the estimating relationship a measure of the technical advance sought in the program, because an analysis of past programs suggests a strong correlation between the degree of advance sought and the subsequent cost growth. However, cost estimating improvements of this type serve only to reduce the unpleasant surprises caused by unanticipated cost growth and do little to improve the acquisition process itself.