Objectives - The purpose of this study is to demonstrate to what extent an evidence based decision model can improve physicians' decisions and whether a selective use of the decision model is feasible. Methods - Four experienced vascular surgeons were asked to make a treatment decision for 137 "paper patient" cases with asymptomatic abdominal aneurysms. Their decisions were compared with the optimal treatment as calculated by a computerised evidence based decision analytical model. Results - Surgeons agreed with the model's advice based on life expectancy in 81% of the cases, and decided to operate in only 12% of the cases for which there was no agreement. Surgeons' decisions differed from the decision model's calculated optimal treatment, in particular for older patients with aneurysms of intermediate size and with many risk factors, and for younger patients with small aneurysms and few risk factors. Not all these decisions, however, were reported to be more difficult. Conclusion - Use of a decision analytical model might lead to more appropriate decisions and a better quality of care. Selective use of the decision tool for difficult decisions only would be more efficient but is not yet feasible because reported decision difficulty is not strongly related to disagreement with the decision tool.