Ride-sourcing services are burgeoning urban mobility services provided by transportation network companies (TNCs), consisting of non-pooling service and ride-pooling service. By receiving online orders and offering efficient door-to-door services, ride-sourcing services are becoming popular alternatives to public transit. Meanwhile, unignorable travel time variability influences passengers’ travel behavior and TNC’s pricing strategy. To investigate the effect of travel time variability, this paper proposes a bi-level framework. The mode choice of passengers, zone choice of searching drivers and the subsequent route choice traffic equilibrium in a transportation network is formulated as the lower level. And the upper level investigates the optimal pricing strategy. In the network formulation, passengers encounter variability of waiting time, detour time and highway travel time. Two pricing strategies are proposed to accommodate the risk of unsuccessful pairing. Pricing strategy 1 charges a fixed price even when a ride-pooling request becomes a non-pooling service due to unsuccessful pairing; whereas pricing strategy 2 increases the price of a ride-pooling request to that of non-pooling service if pairing is unsuccessful. The results show that pricing strategy 1 produces more profit and induces more ride-pooling demand. In addition, under more reliable traffic conditions, the optimal prices for ride-sourcing services are higher and the network travel time decreases. The results illustrate the influence of travel time variability on the ride-sourcing market and provide guidance on the company’s pricing strategy regarding a network under travel time variability.