Introduction. Originality and independence is the first, absolute and general requirement to the content of dissertations, regardless of an academic degree and research direction. At the same time, recently a considerable fraction of different types of plagiarism have been found out in the Russian dissertation works. The aim is an analysis of the most frequent plagiarism in candidate and doctoral dissertations. Results. We consider the established practices in the evaluation of independence in academic works. Unethical use of someone else's work and materials by the authors of dissertations include, in particular, copying from official documents and abstracts without acknowledgement the source and with no citation; replication of existing reviews and chapters from already defended dissertations; falsification by refreshing of outdated data with substituted dates, and or word-for-word copy-pasting with substitution of the research object. The latter type of the plagiarism involving fraud in experimental and statistical data constitutes a special peril for the science and for the society in general. The primary principle for assessment of independence in dissertations has to be proper citing and referencing that should allow a reader to distinguish the author's contribution from someone else's text. Undocumented verbatim quotations in dissertations are inadmissible, whether the author claims scientific novelty or not. At that, it does not matter whether well-known or unique data are provided, and whether the source is protected by a copyright. Practical significance. The qualitative analysis of the factual material, based on the author's observations, can serve as a starting point for the subsequent quantitative analysis of plagiarism in scientific texts.