A new approach to the dynamic layout planning problem is proposed which provides solutions to highly variable material flow patterns occurring over a multi-period planning horizon and is especially suitable for flexible manufacturing systems. A non-deterministic environment is considered in which there is assumed to be uncertainty in the future material flow data. The performance of the method is assessed by comparing the solution produced by this method with a set of data provided in the literature for which the claimed optimal solution is known. There is close agreement with the stated solution and the result is obtained with a fraction of the computational effort. The computational efficiency is due to a new construction method to generate static layout solutions. This method uses an algorithm in which the number of stages is proportional to the number of facilities rather than an exponentional relationship as found in most other methods. The method also uses an element of forward planning to ensure that early location assignments provide minimum restriction to assignments made later in the procedure. Results of extensive tests show that the new static layout planning procedure produces solutions generally better than existing construction techniques and comparable with improvement techniques such as CRAFT. The execution speed of the procedure makes it possible to solve large scale problems ( >30 )in very short time scales on Microcomputers. Incorporation of the fast new construction method into dynamic layout planning allows decision making concerning when and how to re-layout facilities in response to changes in predicted material flow.