Behind all debates over the adequacy of highway revenues lies the tricky issue of how much money states and the federal government ought to spend on highways. States and the federal government have historically tried to determine revenue needs with technical reports known as "needs assessments." These studies usually conclude with a dollar figure that represents the revenue required to bring all roads up to some set of maintenance and performance standards. Even though a great deal of careful technical analysis can go into needs analyses, most do not actually address the question of what total level of spending would be best. Needs assessments typically identify how much money would be required to meet certain standards or to build desired lists of projects, but generally do not address whether or not such standards or lists are optimal. Drawing on examples from California, this paper reviews the evolution of both highway needs studies and fluctuations in highway funding over the past half century. We find, despite efforts to increase the rigor of highway needs analyses, needs studies are often simply "wish lists" of locally popular projects. In particular, cost-benefit analyses have long been proposed to improve the quality and rigor of needs assessments, but have been very slow to be adopted. While a cost-benefit approach to assessing highway needs would inevitably create winners and losers relative to current, engineering and ad hoc-oriented methods of assessing needs, such analyses would provide invaluable information to decision-makers in determining how to spend limited transportation resources most effictively and efficiently.