The authors propose a risks evaluation model for research projects. The model is based on fuzzy inference. The knowledge base for fuzzy process is built with a causal and cognitive map of risks. The map was especially developed for research projects, taken into account their typical lifecycle. The model was applied to an e-testing research project: the probability of not obtaining quality results was computed considering the over-budget sum and the quality level of research idea. The computed risk is situated on the highest level in risks map. A software system for evaluating risks in e-testing research project was also developed. The research was funded by the Ministry of Education and Research, National University Research Council, Grant PCE_ID_873. (Contains 12 figures and 6 tables.)