The paper develops holonic architecture for reconfigurable manufacturing systems (RMS), which are capable of adapting to unpredictable changes in demands. RMS are designed to produce various products grouped into families in a short time at low cost.Aholonic structure reflecting basic holons forRMSis developed and then linked to an analytical network process (ANP) model, as a multi-criteria approach, to evaluate system performance. The proposed generic model provides flexibility for holons and facilitates evaluation of RMS considering economical and operational aspects as the main performance objectives. Moreover, new requirements such as functionality and capacity for process reconfigurability along with reconfiguration time/cost are taken into account. By allowing interactions among all the ANP clusters and their relevant elements in terms of outer and inner dependencies, the critical factors affecting the system performance are explored and evaluated through a case study. In particular, the criticality of the elements affecting the system performance will be assessed with respect to planning horizons, economical/operational aspects, and process reconfigurability based on available capacity and feasible functionality.