Abstract The conventional approaches to PPC are being reconsidered with respect to the shift in production goals from maximum output at minimum cost to high flexibility, quality, reduced throughput-time and high schedule observance. On the level of workshop control the implementation of a generated schedule in the shop is impeded by unavoidable disruptions. Since process planning is usually performed a long time before manufacturing, the load situation and the availability machines in the shop is not taken into account. Investigations have shown that in many small and medium-sized metal-working companies, 20 to 30% of the total load has to be redirected to other resources and that only a small amount of workshop orders actually complies with the original short-term schedule. This situation can be improved significantly by using flexible process plans that comprise manufacturing alternatives. These flexible or non-linear process plans allow to reconsider the final manufacturing route only shortly before or during the workshop scheduling process, when the actual situation in the workshop is known. On the other hand it facilitates to reschedule orders to other manufacturing resources in case of unforeseen disruptions. This paper describes developments achieved within the ESPRIT Project 2457 FLEXPLAN. FLEXPLAN develops an integrated process planning and scheduling system based on ‘non-linear’ process plans (NLPPs). These process plans are represented as Petri nets that resemble a net of feasible operation sequences. The system comprises knowledge-based automatic process planning functions for a limited spectrum of workpieces, a graphical editor for non-linear process plans and an hierarchical workshop scheduling system that utilises the NLPPs to increase flexibility in the workshop. The workshop scheduling system comprises a load-oriented module for medium range scheduling, a finite scheduling module with graphical user interface and schedule evaluation functions. The optimisation algorithm used in the finite scheduling is based on a rule-based execution of petri-nets.