AbstractThe choice of an audit sample involves integration of diverse goals. Such integration must consider the preferences of the auditor towards costs as well as risks. Traditionally this problem has been formulated to determine minimal sample size and its allocation such that each of the goals is satisfied simultaneously or to the extent possible. However, these formulations do not consider trade-offs between goals, so that the sample may not reflect the auditor's preferences regarding sampling costs and risks. In this paper, we model such trade-offs by utilizing a polynomial goal-programming framework and utilize a practical case problem to illustrate the framework. Computational experience shows the technique to be feasible and inexpensive.