The indications for many laboratory tests in patients with uveitis are controversial. Bayes' theorem allows a mathematical approach to the assessment of the utility of a laboratory test based on the sensitivity of the test, the specificity of the test, and the pretest likelihood that the disease the test is intended to identify is present. We have utilized Bayes' theorem to assess the utility of routine antinuclear antibody and purified protein derivative testing in patients with uveitis. Based on published data about the sensitivity and specificity of each of these tests, as well as the prevalence of systemic lupus erythematosus and tuberculosis among patients with uveitis, we calculated that a patient with uveitis and a positive antinuclear antibody test result has less than a 1% chance of having systemic lupus erythematosus and that a patient with uveitis and a positive purified protein derivative test result has a 1% likelihood of having tuberculosis. These low probabilities mean that neither test is useful in the routine evaluation of patients with uveitis, and indiscriminate use may lead to improper diagnosis, increased costs, and, occasionally, inappropriate therapy.