This paper proposes a DEA-based model for analyze the fuzzy risk in project selection. We used concept semi-variance for measure upper and downside risk and a DEA model for Classification desirable and undesirable risk. Firstly, the proposed model includes new desirable and undesirable risk-return indexes. Thus a novel DEA model is presented for evaluation and Classification desirable and undesirable risks and finally, is extend to fuzzy DEA model for project portfolio selection. An applied example is used to explain the proposed approach and usefulness and applicability of this approach have been illustrated using the 37 available projects. Manuscript profile