The purpose of the presented work is to develop an approach that allows reducing the degree of subjectivity in assessing the complexity of a test task. The educational process consists of many different disciplines. Disciplines differ in the content of educational material, volume, number of hours devoted to study, type of control and tasks. Accordingly, teachers who make up questions of varying degrees of complexity (simple, medium complexity, increased complexity), when setting the complexity, may introduce some subjectivity. To obtain a timely assessment of the level of knowledge of trainees, a verified bank of test tasks of varying complexity is required. Adaptive testing allows to get an adequate assessment of trainees quickly. Adaptive testing is carried out using tasks of varying complexity. It is necessary to have a formal apparatus, that allows to get a bank of tasks with varying degrees of complexity. A formal apparatus reduces the degree of subjectivity in assessing the complexity of tasks. A bank of test items with calibrated questions with varying degrees of complexity will allow for adaptive testing in a number of disciplines. The solution of the problem posed is difficult to perform with the available classical methods of information processing. This problem can be solved using the apparatus of fuzzy sets, which allows to process parameters of various nature.