The degree of possibility is used by the extent analysis (EA) to extract the crisp local weights from a triangular fuzzy comparison matrix (FCM); while, it is just a relation for comparing two triangular fuzzy numbers. This causes EA to extract the inaccurate local weights from a FCM and sometimes to rank the alternatives incorrectly. In this study, we revise the EA to avoid this misapplication. For this purpose, we propose a procedure in which the global weights of alternatives are obtained as fuzzy values; then, the degree of possibility is used just for ranking the alternatives based on their fuzzy global weights. The advantages of the revised extent analysis (REA), proposed in this study, compared with other fuzzy analytic hierarchy process are its simplicity and low computations. This study gives three numerical examples to illustrate how REA avoids the flaws of EA.