In recent years, a new efficient delivery service mode in city area has developed rapidly, that is, to realize the requests delivery service with delivery constraints from point to point on the same day for stochastic requests. Aiming at the requests delivery problem with delivery constraints and stochastic, considering the features that the platform can either accept the requests of its vehicle or entrust the requests of the third-party logistics, a requests dispatch strategy is proposed to decide whether to accept the requests and determine the vehicle service requests set. A Markov decision model for requests delivery is established to minimize the operating cost of the platform. Then, the properties of the optimal strategy are analyzed. Finally, the results show that compared with the FCFS strategy and on-time delivery strategy, the order distribution strategy proposed can reduce the total operating cost of the platform and improve customer satisfaction and ensure service quality. The effectiveness of the model and algorithm is verified by a set of numerical analyses.