THE OBJECT IN SUCH A CASE IS ALSO TO DETERMINE THE OPTIMUM POINT AT WHICH BENEFITS ARE AT A MAXIMUM. THE COSTS AND BENEFITS PERSPECTIVES ARE ALGEBRAICALLY IDENTICAL SINCE COSTS ARE NEGATIVE BENEFITS AND BENEFITS ARE NEGATIVE COSTS. THE TWO MUST BE CONSIDERED TOGETHER. THE PAPER IS ALSO CONCERNED WITH THE METHODOLOGY OF DEVELOPING CAUSAL MODELS WHEREBY BOTH THE EFFECTS ON VARIOUS COSTS OF CHANGING THE POLICY LEVEL AND THE EFFECTS OF COST CHANGES ON THE OPTIMUM POLICY LEVEL ARE SHOWN. SPECIFICALLY, THE EXAMPLE OF PRETRIAL RELEASE IS USED; SOME OF THE CONSIDERATIONS APPROPRIATE TO DETERMINING AN OPTIMUM PERCENTAGE OF DEFENDANTS TO RELEASE PENDING TRIAL ARE DISCUSSED IN LIGHT OF VARIOUS GOALS AND ENVIRONMENTAL CONDITIONS. IMPORTANT FACTORS TO BE CONSIDERED INCLUDE EXCESSIVE JAIL COSTS, LOST GROSS NATIONAL PRODUCT, AND BITTERNESS COSTS WHEN THE PERCENTAGE OF DEFENDANTS HELD IS TOO HIGH; WHEN IT IS TOO LOW, THERE ARE EXCESSIVE COSTS IN REARRESTING THOSE WHO DO NOT SHOW UP FOR TRIAL AND EXCESSIVE COSTS TO SOCIETY DUE TO CRIMES COMMITED BY THOSE WHO SHOULD HAVE BEEN KEPT IN JAIL. THE VARIABLES USED IN THIS STUDY ARE EXPLAINED. CAUSAL MODELS ARE CONSTRUCTED SHOWING THE EFFECTS OF INCREASING THE PERCENTAGE OF DEFENDANTS HELD AND THE EFFECTS OF COST CHANGES ON THE OPTIMUM PERCENTAGE OF DEFENDANTS TO HOLD. ALTERNATIVE MODELS ARE ALSO EVALUATED. THE METHODOLOGY OF FINDING THE OPTIMUM PERCENTAGE TO HOLD IS DETAILED, AND EXPANDED APPLICATIONS OF THE METHODOLOGY TO OTHER POLICY PROBLEMS ARE DISCUSSED. A GLOSSARY OF TERMS IS INCLUDED, AND AN APPENDIX LISTS BASIC FORMULAS USED.