Physicians must frequently combine statistical information on prevalence of diseases and on medical tests according to Bayes’ theorem in order to arrive at a correct diagnosis. Such decisionmaking processes are often associated with significant errors of judgment, which have been documented repeatedly with respect to the Bayesian “standard” task (i.e., one disease must be diagnosed based on one medical test). In the present contribution,we generalize the Bayesian reasoning paradigm to medical situations where more than just one medical test is involved and suggest three steps towards a better understanding of that generalized situation, namely, 1) replace probabilities with natural frequencies; 2) present a tree diagram containing natural frequencies; and 3) highlight the two branches of the tree that are relevant for the requested inference.