OBJECTIVES: To develop a utility-maximisation model of choice between alternative treatments of specific disorders on the basis of microeconomic consumption theory. The model is to take into account patients´ preferences for clinical procedures, possible outcomes and risks involved in alternative treatments. METHODS: The optimisation problem facing a patient is presented by geometrical and algebraic analysis using an ordinal von Neumann Morgenstern (vNM) utility function in a simple example involving two treatment alternatives, T1 and T2. As the treatments are perfect substitutes, the ratio of marginal utilities (dT2/dT1) which determines the slope of a patient´s indifference curves is constant and can be identified by means of a modified standard gamble (MSG). By including relative prices of interventions, optimal allocation decisions are derived, hypothetically assuming a market context. RESULTS: vNM utility functions are developed for the alternative treatments and a "no treatment" option (T0) on the basis of outcome and probability data from clinical research. Utilities of procedures and outcomes can be identified empirically. This requires a preference ordering of treatment alternatives and MSG questions relating to hypothetical treatments. The number of MSG questions and hypothetical treatments (n-2) depends on the number of vNM variables (n). This procedure identifies the ratio of marginal utilities of T1 and T2, dT2/dT1. The inclusion of relative prices of T1 and T2, pT1/pT2, into the model allows the prediction of utility-based treatment choices and the application of the model in resource allocation. CONCLUSIONS: The main advantages of this model are its ability to predict utility-based decisions incorporating risk preferences and to avoid the need for methodologically delicate techniques such as sensitivity analysis and discounting. The practical benefits of this approach remain to be determined within empirical evaluations.