SUMMARY
In this paper the theoretical model of self-learning based on the use of probability of incorrect assignments of the test for a computer program for monitoring the implementation of tests as the open form is suggested. The results of simulation modeling and pedagogical experiments conducted by using the developed program which implement the self-learning algorithm, are showed the validity of the theoretical model and the appropriateness using of self-learning control programs for differentiation efforts of students on the realization of these tests. The article’s submissions may be used for further studies of the self-training process of software that realize other types of tests and test development of knowledge control and ability based on them.