The purpose of this article is to provide an understanding of methods that are useful in formulating advice about when to use diagnostic tests. If the clinician expresses diagnostic uncertainty as the probability of a disease in a patient, Bayes' theorem may be used to predict the effect of doing various tests and their impact on patient management. To use Bayes' theorem wisely, one must be aware of pitfalls in estimating probability and must understand the limitations of most studies of the sensitivity and specificity of diagnostic tests.