Each reform, to varying degrees and in different ways, was based on erroneous assumptions and 'crisis thinking' and, as a consequence, was overextended because of exaggerated claims and the lack of careful evaluation of test projects. Failure is often rooted in fundamental misunderstandings of the problem, the processes of change, and the dynamics of the system, as well as lack of attention to detail when delivering services. Fragmentation and adaptation (the tendency to subvert reforms by reintroducing discretion at a different stage of the process) are the central and continuing obstacles to change in the criminal justice system. Each of the four examples relied on an administrative strategy for coping with these obstacles. A problem-oriented strategy would insist on a realistic appraisal of problems facing the court and on sensitivity to administrative details. Litigation, research and development, and modest national support for experiments in the courts can provide a basis for achievable reforms. Reference notes and an index are appended.