The operational level of flexible manufacturingsystems (FMS) is concerned with the decisionmaking required for real-time operation. The real-time controlof FMS is extremely complex. On-line scheduling isreferred to as real-time control of FMS. On-line schedulingof an FMS requires decision making in various schedulingproblems such as selection of an AGV and a workcentre from a set of workcentres simultaneously requesting the service fortransport of a part, selection of a new part type to be releasedinto the system, etc. This paper describes a prototype knowledge-based system for selection of an AGV and a workcentrefrom a set of workcentres simultaneously requesting theservice for transport of a part in on-line scheduling of FMS.The knowledge-based system is evaluated by the empiricalapproach. On-line scheduling of FMS gives potential advantagessuch as productivity and low throughput time. © 1997 Taylor & Francis Group, LLC.