Parallel processing comprises the concurrent use of multiple processors to speed execution of one operation. Although techniques suited to most of the common geographical data models have been prototyped, the prominent exception has been vector-topology. This paper explores whether operations that create a vector-topological dataset can benefit from parallelisation. It describes techniques for using multiple processors concurrently to create vector-topology for multiple sub-areas, and for stitching these sub-areas together to form the resultant dataset. To achieve performance gains over sequential processing, the overhead of the stitching must be less than the gains from the parallel processing of sub-areas. These methods are tested in the context of the conversion of raster data to polygonal vector-topology. Speed-up in comparison to single-processor performance is achieved on both a 4-processor shared-memory Sun server and using up to 15 processors of a Cray T3E. The approach taken hides the parallelism and the management of the vector-topology in a software framework that simplifies the task of parallel application development.