Clinical use of a test should be evaluated not only on the change of pretest probability but also on the increased confidence in the judged probability, and on the decision of recommending additional tests or therapy after test results. Before and after the test, cardiologists referring a patient for exercise electrocardiogram for suspected coronary artery disease were asked to estimate the probability of coronary artery disease and the minimal and maximal value of their estimate (plausible range), to judge whether to refer the patient for coronary angiography, and to specify patient's treatment. After the test, the percentage of patients with estimated probability < or = 20% increased (from 53 to 69%; p = 0.005) and the percentage of patients with intermediate probability (between 21 and 80%) decreased (from 37 to 16%; p = 0.0001). The mean plausible range decreased from pre- to post-test assessment (31 +/- 23 to 15 +/- 16%; p = 0.0001), indicating an increase in cardiologists' confidence in their estimates. After the test, the percentage of cardiologists who wished to either refer or not refer patients for coronary angiography increased (from 4 to 13%, p < 0.005; and from 37 to 65%, p < 0.0001, respectively). Test results enable cardiologists to change the assessment of disease probability, the confidence in their own estimates, and their judgment of whether to refer patients for coronary angiography.