In order to generate a driving cycle for evaluation of vehicles emissions, extensive emission data and driving activity data are collected by using a Portable Emission Measurement System (PEMS) and a GPS device.The collected data are preprocessed and the evaluation criterion for microtrips is proposed.Two heuristic algorithms, Genetic Algorithm (GA) and Discrete Simultaneous Perturbation Stochastic Approximation (DSPSA) are given to search the best microtrips automatically.The prediction of vehicle emissions using generated driving cycle is validated.Research results show that both heuristic algorithms are able to solve the problem of automatic generation of driving cycles.The driving cycle developed by GA is more representable and it can predict the total vehicle emissions with a relatively small error.Therefore, GA is recommended for the generation of driving cycles in other cities.