An analytic procedure, efficiency analysis, is proposed for improving the utility of quantitative program evaluation for decision making. The three features of the procedure are explained: (1) for statistical control, it adopts and extends the regression-discontinuity design; (2) for statistical inferences, it de-emphasizes hypothesis testing in favor of interval estimation; and (3) it uses the limits of the confidence interval to qualify the level at which a program operates, rather than making a simple statement about goal attainment. Application of this procedure is illustrated with data obtained at grades 2, 3, 7, and 8 for a remedial mathematics program. Methods are suggested for making these evaluation results accessible and practical for educational decisionmakers through graphic presentations and by linking evaluation outcomes to particular administrative objectives, specific planning procedures, and a set of corrective and/or supportive program activities. (Author/GDC)