THE AUTHOR CONTENDS THAT PRESENT PROCEDURES FOR COMPLETING EVALUATIONS OF DELINQUENCY PREVENTION PROGRAMS SUPPORT OUTDATED ASSUMPTIONS (FOR EXAMPLE, THAT INSTITUTIONAL STATISTICS ARE EVIDENCE OF PROGRAM EFFECTIVENESS, OR THAT JUVENILES MUST BE 'TREATED'). HE ARGUES THAT THESE ASSUMPTIONS TEND TO MAKE IT DIFFICULT TO LEAP OVER CURRENT OBSTACLES, RECOGNIZE DIFFERENT REALITIES, AND CONSIDER ALTERNATIVE OPTIONS TO PRESENT ACTIVITIES. SECONDLY, THE AUTHOR MAINTAINS THAT EVALUATION PROCEDURES TEND TO GUT INNOVATIVE PROGRAMS DO NOT HAVE THE SUPPORT OF THE PUBLIC OR OF THOSE ESTABLISHED IN 'CREDITABLE' AGENCIES OR PROFESSIONS. FINALLY, IT IS ARGUED THAT CURRENT APPROACHES TO EVALUATION HAVE LITTLE RELEVANCE FOR GOVERNMENT POLICY. THE AUTHOR CONCLUDES THAT GOOD RECORD KEEPING AND CAREFUL ANALYSIS OF STATISTICS MIGHT BE A MORE SUCCESSFUL AND PROMISING METHOD OF ASSESSING PROGRAMS THAN PROGRAM EVALUATIONS.