The author describes five types of crime prevention programs whose rationale, design, and execution seemed promising, but which caused harmful effects. The analyses of the programs show how social programs can contribute to crime as well as reduce it; they can also increase illness and reduce the ability of clients to cope with life challenges. Evidence on two types of programs assessed--those that involve court volunteers and those that provide healthful group activities--have appealing research designs, but no evaluation results have been published. This reflects a strong bias against reporting the adverse effects of well-meaning social programs. Even when independent evaluators attempt to publish adverse findings about a program, editors may refuse to publish the results because the findings are perceived as having no constructive value. Another problem is that inadequate evaluations, notably those that rely on clients' expressions of appreciation, fail to detect adverse program effects. Some programs may prevent some types of crime but promote other types, and some may reduce crime while contributing to mental illness and alcohol abuse. The harmful effects of a program can only be detected when a comparison control group is used. If the treatment group shows more adverse outcomes than the group that received no treatment, then participants would have been better off not to have been in the program. Also, evaluations should assess a broad range of variables. If the evaluation focuses too narrowly on only a few outcome measures, it may miss some aspects of participants' lives that were adversely affected by the program. 29 references