In the paper the complex approach to scientific and technical document quality assessment is proposed based on various automatically calculated document quality characteristics as widely used bibliometric and scientometric (based on citation indices), and the new types of characteristics based on the text semantic analysis, heuristics, and also on plagiarism detection methods. The integrated indicator of scientific and technical document quality assessment is formed on the basis of the received basic characteristics with use of machine learning methods similar to the problem of ranking in information retrieval. The developed prototype system based on offered approach is presented, and also the experimental investigations of the developed system directed on check of scientific and technical document quality assessment accuracy are carried out. The analysis of the state of art researches of scientific and technical document quality assessment showed the offered approach based on enhanced list of basic characteristic groups was considered by nobody in so broad statement and as a whole is innovative. The main part of the paper has the following structure. The second section contains an analytical overview of existing approaches to assess quality of scientific and technical documents. The third section provides detail of a proposed approach to assess quality of scientific and technical documents. The forth section describes a prototype system based on the proposed approach. The fifth section discusses results of experiments.