We study the optimal pricing decisions of a ride-hailing platform that offers a probabilistic service, in which a customer can hail multiple types of ride services simultaneously and has a chance of obtaining one of them. We develop a stylized model that captures the heterogeneity of customers’ preferences and the different service qualities of the ride services. We consider two pricing schemes: allocation-dependent pricing, where the customer pays according to the service she receives, and dedicated pricing, where the customer pays a fixed price for each type of service. We characterize the customers’ equilibrium choice behavior and the platform’s optimal pricing decisions under each scheme. We find that introducing probabilistic service increases the platform’s revenue and improves the service efficiency and quality, by achieving the pooling effect of the service facilities and a finer customer segmentation. We paradoxically find that, with the probabilistic service, a larger service capacity may reduce the platform’s revenue. We also find allocation-dependent pricing can lead to a higher average consumer utility comparing to dedicated pricing, while dedicated pricing further enhances the platform’s revenue, by allowing the platform to charge a separate price for the probabilistic service.