The major objectives of this thesis are as follows. (1) In the ex-post evaluation of R\&D projects, many qualitative evaluation items are used but the reliability of these items is questioned. So the reliability of these items is inspected in this thesis. (2) For the suggestion of reliable items, the reliability of evaluation items is investigated considering the types of institutes and projects. (3) For the improvement of evaluation quality, Influencing factors of the reliability of evaluation are investigated. The unit of analysis is a project in the institutes of large size firms and government supported institutes. To investigate the reliability of the qualitative items, intraclass correlation coefficient is used. The major findings of this thesis are as follows. (1) In the ex-post evaluation of R\&D projects, the reliability and validity of evaluation are increased by subdivision of qualitative evaluation items. (2) Considering the types of institutes, items about commercial success and technical practicalness are reliable in the private institutes but items except possibility of patent acquisition are not reliable in the government supported institutes. (3) Considering the types of projects, items about commercial success, technical practicalness, strategic necessity and possibility of patent acquisition are reliable in the development projects but only two items about possibility of patent acquisition and meeting assigned schedule are reliable in the research projects. (4) Considering project performance as one of influencing factors of reliability of evaluation, project performance is lower, evaluation result is more reliable.