A set of formulas for the indices of the performance and predictive ability of the binary genetic tests is presented. Their dependence on disease prevalence and the population frequency of a genetic marker is characterized. It is shown that a marker with the odds ratio OR < 2.2 has an initially low prognostic efficiency in every sense and at any frequencies of the disease and the marker. A marker can be a good classifier, when OR > 5.4, but only in case its population frequency is rather high (p _M > 0.3). The formulas are presented that allow obtaining indirect estimates of the absolute and relative risk of the disease for the carrier of a marker in the case-control studies.