The increasing number of complementary examinations in pneumology often makes it difficult to choose the appropriate tests for optimal investigation. After specifying the conditions in which reference values are established, the authors discuss the general principles of decisional analysis, i.e. sensitivity and specificity, probability ratios, the role of the disease prevalence in interpreting results (Bayes' theorem) and the odds ratio. Practical applications of these principles with regard to diagnosis and therapy in pneumology are also discussed.