Assuring the quality of an evaluation is a great challenge to evaluators. The evaluation of an evaluative process is called meta- evaluation. In Brazil there is a great concern about evaluation quality, although the concept of metaevaluation is new. Evaluation professionals are still discussing the criteria of excellence that a true evaluation should attend. The standards defined by the Joint Committee on Standards for Educational Evaluation (1994) that are presented in terms of utility, feasibility, ethics and accuracy, enlighten on that discussion. This study presents a new methodology for meta-evaluation that makes use of fuzzy sets and fuzzy logic concepts. The methodology is composed of an instrument for data collection (checklist for the meta-evaluation of programs/projects) and of a hierarchical fuzzy inference system for treatment of data related to the metaevaluation of projects and programs. The main advantages of using such a system are: (i) the possibility of working with linguistic rules; (ii) the use of tools that deal with the intrinsic imprecision of complex problems, as is the case of metaevaluation; (iii) the incorporation of subjective knowledge (of specialists); (iv) the adaptability of the inference process to specific situations. This new methodology makes use of a system composed of 36 rule bases organized in three levels: standard (level 1), category (level 2), and meta- evaluation (level 3). The main features of the proposed methodology are: (i) the data collection instrument, that allows intermediate answers; (ii) the fuzzy system's capability of adaptation to specific needs; (iii) the transparency provided by the use of linguistic rules, which favours understanding and discussion of the whole process. It is believed that this methodology will make the meta-evaluation process easier. This study intends to be a contribution to evaluation as a subject and to meta- evaluation practice.