The purpose of the research was to identify and study the features of the effect of four questions forms and their combination in polymorphic test on the output indicators of secondary school students’ testing. In the study, conducted in the subject field of studying native (Ukrainian) language, took part students of 8th and 9th-grades (N=164). The tools (criterion-oriented author's tests, questionnaire) and methods of results’ analysis (factor analysis, dispersion analysis, etc.) used in the research process, made it possible to prove the existence of the effect of factor "question form" on the test scores (impact power is up to 35%), and also to determine combination of question forms that is characterized by the highest diagnostic accuracy. Results of a study indicate that it is advisable to construct a polymorphic linguistic test. The combination of the multiple-choice questions with one correct answer with using the text and constructed-response questions can ensure the clarity of output indicators of testing (with the least effect of question forms on them), in case evenly balance of these questions in the test. The disadvantages of the study are pointed out and prospects for their continuation and deepening, connected with further study of the optimal empirically valid test structure, are substantiated.