The dependency of the predictive values of tests on the prevalence of diseases is an escrow issue of clinical diagnostics. The relation between pre-test probability and post-test probability is well explained by Bayes' theorem, and the relation between prevalence and false diagnoses can be described well by modifying this theorem. In cases of low prevalence the positive predictive value (PPV) is lower and the false-positive predictive value (FPPV) higher. These aspects mainly depend on the test specificity. But basically, in cases of low prevalence there is a higher negative predictive value (NPV) and a lower false negative predictive value (FNVP). Depending on the sensitivity and specificity, NPV and FNPV vary only slightly in low prevalence ranges. These statistical relations are able to explain the typical mode of operation of general practitioners with their unselected patients. In order to increase PPV and decrease FPPV in the diagnostic workup, the GP must use his clinical experience, time and stepwise diagnostic procedures. More diagnostic studies are necessary to improve the diagnostic workup in patient care.