Industrial performance can be defined in terms of numerous criteria to be synthesized for global control purposes, many of them being of a complex nature, i.e. not related to one elementary physical measure. In this context, determining a global performance expression raises the issue of performance expression aggregation. To address such an aggregation issue, so-called performance measurement systems need to be implemented. The analysis of the corresponding literature leads to the conclusion that the majority of the proposed approaches either do not provide explicit aggregation mechanisms or propose overly simplistic methods really to cope with the complexity of the situations at hand. Only a few consider explicit adequate aggregation methods. Therefore, a thorough characterization of performance expression aggregation based on four stages, namely extraction, representation, combination and interpretation, as proposed by the information aggregation community is presented. The contribution deals with the definition of a performance expression combination based on two kinds of performance expressions commonly encountered (physical measures and performance evaluations, i.e. objective satisfaction degrees). Methodological guidelines for performance expression aggregation are proposed as well as associated mathematical tools, especially the fuzzy Choquet integral for taking criteria interactions into account. An industrial case study is used as an illustration, and some concluding remarks and emerging problems to be considered in the future, such as temporal aggregation, are finally pointed out.