The relevance of this study is determined by the algorithm developed for test task selection. The purpose of this article is to develop this test task selection algorithm having single and multiple-choice answers with various blocks of reactions. The main approach to the study of the problem is the construction of a mathematical model to calibrate test task parameters and the university students’ scale of knowledge level. The study has proven insufficient use of the classical theory of testing in an objective assessment of students’ knowledge. The method for calibrating test task parameters is developed. Scales of testees’ readiness level are defined. The adequate size of test reliability coefficient is found. The rule for test task selection is formulated. A formula for the complexity degree of a test task is found. An algorithm for allocation of test task types with unambiguous and multiple-choice answers with various blocks of reactions is offered.