The author examines the issue of publications selection for the research reviews at graduate schools. Despite the significance of review papers for systematization of findings and the state of knowledge, assignment of the review is often offered to students without sufficient specification of review type, tasks and procedures of working with primary sources. As a result, learners are not aware of how to accumulate literature corpus and pass from discussion of individual publications to analytical generalization. The purpose of the article is to substantiate the relations between the review type and requirements to publications selection, and to define pedagogical environment for building relevant research skills. Theoretically, the paper is based on foreign typologies of review studies and the national tradition of information analytic and synthetic processing. Through comparing these foundations, the author links review type to search, selection and generalization characteristics. The author introduces the conceptual model incorporating review type, source selection, generalization method and learning outcomes, as well as the matrix of pedagogical approaches to design of learning assignments and review paper evaluation. She suggests to organize the leaning process according to the successive complication: from search, selection and first analysis of publication to comparison of study approaches, and further - to formalized selection procedure and search strategy description. The originality of the paper resides in considering review practices as a stage-by-stage pedagogical process linking review method with the level of research skills. The author demonstrates that the completeness, substantiation, and scientific value of a review paper is determined largely the quality of literature selection.