The aim of the research work was to analyse the accuracy (and reproducibility and resolution) of passive marker-based motion analysers. We elaborated an ap-propriate model and measurement set-up. Based on the results we developed such image processing algorithms that are applicable in simple two dimensional motion analysers. Many applications would require the replacement of human observations with automatic motion analysers but only a few can afford the cost. Our results greatly help for the development of motion analysers using commercial video cameras. In medical applications the motion analysers must operate autonomously. The research work comprised the analysis of algorithms objectively qualifying the actual state of patients with neurological disease (especially Parkinson's disease).