The authors argue that high-quality evaluation designs -- for example, comparing experimental and control areas and including before and after measures of crime -- are needed to evaluate situational crime prevention initiatives. In a previous design of this kind that used household victimization surveys to measure crime, the authors showed that crime decreased after street lighting was improved. The current study examined whether the same results were obtained in a self-report survey of young people, which was also conducted in experimental and control areas before and after the improved street lighting. The authors argue that self-reported delinquency is a valid and reliable measure of offending. The self-report results corroborated the victimization survey results in showing that offending decreased in the experimental areas compared to the control area. Also, the responding youth thought that the crime problem had decreased more in the experimental area, and their fear of crime after dark also decreased more in the experimental area; however, the victimization of youth did not decrease more in the experimental area, possibly because street pestering by older people did not decrease. 3 tables, 27 references, and appended discussion of comparability of groups