Within the RERAU methodology, each rehabilitation criterion is assigned a grade out of 4 possible ones. This grade results from successive aggregations of performance indicators. Issues related to base data (uncertain or imprecise data) and to fusion (conflicting or reinforcing indicators) lead to envisage a fuzzy method for processing the proposed set of performance indicators. Five issues are discussed in this paper: representing an indicator with fuzzy linguistic variables; representing each aggregation table with “if... then...” rules; combining two premises of a rule; fuzzy representation of the implication; aggregating the results of several rules.