Petri net and supervisory control theories are popular approaches used to model the control of automated manufacturing systems. However, these methods do not adequately resolve the following three critical issues in modeling the control of a flexible manufacturing system (FMS) on the shop floor: flexibility, complexity, and applicability. This paper presents a new formal design methodology, called structured adaptive supervisory control (SASC), for constructing the control model of an FMS by resolving these issues. The SASC model of an FMS takes into account all the legal control specifications in terms of system capability, rather than the control specification in terms of a given part mix within an FMS (as is the case with Petri net and supervisory control models), to meet the needs of system flexibility. Based on the theory of finite capacity machines, an SASC model whose complexity growth-function is linear in the size of an FMS then can be formed. To improve the model-construction efficacy and make a design economic and applicable, the SASC model is hierarchically structured as a well-defined module with three control layers: an acceptor, an adaptive supervisor, and an executor. The acceptor checks on the capability of completing an incoming task; the adaptive supervisor dynamically resolves all the interactions and coordination between equipment, thus guaranteeing successful completion of all the received tasks; the executor performs in real time all the physical operations under supervision.