Changes in criminal codes can take many forms; they can be directed at an enormous variety of problems and implemented through a wide range of mechanisms. Consequently, there is no single set of issues and no single methodology for evaluations. This study focuses on the particular difficulties associated with internal validity and properly characterizing the treatment. It is argued that effective legislative evaluation must answer three basic questions: (1) Does the content of the legislation matches the legislators' objectives? (2) Is the legislation properly implemented? and (3) What is the impact of legislation on targeted populations? An example is legislation on family violence, the purpose of which is to increase the probability that offenders will be severely punished. However, if the legislation is poorly drafted (i.e., inconsistent with sound legislative theory), it will have the opposite effect of lowering the probability of severe punishment for this type of offenders, because prosecutors will hesitate to prosecute any but the most severe cases in the realization that convictions will be hard to obtain. Another example is legislation aimed at securing an immediate, if modest, payment of unemployment benefits to released ex-offenders until they find work. If bureaucratic procedures prevent processing of the ex-offenders' unemployment claims in less than 4 weeks, the legislation will not be properly implemented, and its objective of affecting recidivism may not be achieved. Finally, evaluators must determine the impact of the legislation on the targeted outcomes, which is the usual goal of all evaluation research. A causal perspective for evaluations of criminal justice legislation is illustrated in a figure. Endnotes and 24 references are appended. (Author abstract modified)