This study has investigated the performance of some well known microscopic traffic flow concepts based on differential GPS data. Several models are calibrated using genetic algorithm and compared on the basis of how good the models fit with the experimental data. The General Motors' model is perhaps the most well know car following model, however there exist several contradictory findings on its parameters. The GM model is calibrated using two different approaches to identify more reliable parameters. The models mostly performed closely and fitted well with speed data compared to headway data. The interpersonal variations are found relatively higher than the inter-model variations. The GM model parameters are identified in a narrow range with R2 values significantly improved.