The growing complexity of systems and the need for executing large projects have led to the development of complex fl exible manufacturing systems (FMS) demanding specifi c control-monitoring architectures. The problem of failure occurrence tends to increase according to this complexity leading to time-consuming tasks as the localisation and repairing. The occurrence of failures during the exploitation stage can deeply modify the FMS performances or its availability. In this context, the maintenance integration into a control-monitoring system becomes an important issue, improving production time and minimizing unplanned costly breakdowns of FMS. We want to investigate the problem of triggering the maintenance, giving a useful decision support tool to evaluate the system availability since the control system’s early design stage. It also results in improvement of the system’s functionality in terms of effi ciency, productivity and quality. This paper proposes a control- monitoringmaintenance architecture (CMM) for FMS based on Petri nets with objects (PNO), where stochastic rates are associated to the modelling of maintenance planning. This framework is based on a modular and hierarchic model structured in CMM modules. The integration is based on a development methodology in which the maintenance aspects and policies are taken into account from the conception (modelling) stage. These efforts acts as a basis for the control architecture of a robot-driven fl exible cell, connected to the Ethernet-TCP/IP.