An important decision-making problem raised in diagnostic systems is the selection of an appropriate combination of tests that can improve the accuracy of the diagnostic system at lower examination costs. To address this, the present study established general performance indices for diagnostic tests and diagnostic rules for test combinations. Based on these, non-dominated test combinations were derived using a multi-objective optimization method, and preferred test combinations were selected by examining the relationships among performance indices, costs, and utility through prescriptive cost-effectiveness analysis. When this analytical method was applied to a database of short-term laboratory tests, the results demonstrated its effectiveness in preventing unnecessary tests and the dysfunctional side effects caused by test administration, as well as in enhancing the utility of examinations.