Following a history of voluntary guidelines development, the paper discusses two broad sets of issues: the appropriateness of empirically based methods in guidelines development and the correctness of the statistical procedures employed. An account of Philadelphia's Bail-decisionmaking Project emphasizes that bail decisions under guidelines revealed clear-cut differences compared with a nonguidelines control group, particularly in disparity reduction. Analysis of criticisms that modeling practices may generate biased decisionmaking concludes that the impact of poor modeling procedures may be much less than fear. The paper finds insufficient evidence to discredit claims of the decision guidelines approach to policy reform. Tables and approximately 30 references. (Author abstract modified)