This article proposes two optimization models to solve the multi-satellite on-orbit hybrid service mission planning problem under uncertainty. During the process of on-orbit refueling and debris removal, two categories of uncertainties are taken into account. These uncertainties have an impact on the mass of the service satellite as well as its fuel consumption during the orbit transfer. The two-stage orbit transfer strategy is given to save fuel for maneuvering, which also determines the directed graph between target orbits of satellites or debris. The stochastic expectation model is to get the service sequence with the optimal average benefit, while the robust model considers the worst situation and gives the optimal minimum benefit. Simulations of GEO satellites in orbit are conducted to demonstrate the impact of various models on handling uncertain tasks.