Jupiter’s icy moon Europa is among the most promising locations to explore for signs of extraterrestrial life in our solar system. However, searching for biosignatures on the surface of Europa presents an unprecedented set of challenges: immense uncertainty, limited energy, and few opportunities for operator feedback. Effectively carrying out a robotic science campaign under these conditions will require a system with a greater degree of autonomy than any planetary exploration mission to date. In particular, onboard scheduling and execution are required for robustness to uncertainty in surface conditions, variation in lander performance, and science discoveries to maximize the quantity and quality of science data returned to Earth during a fixed, limited lander lifetime. Here we represent a proposed Europa Lander surface mission as a utility-driven hierarchical task network and establish through analysis that onboard autonomy using automated mission planning with execution feedback and predictive task models results in mission execution that is more consistently productive compared to traditional static approaches. We design a simulated onboard autonomy framework built by integrating two software components—Multi-mission EXECutive and Europa Lander Autonomy Prototype—to properly simulate the Europa Lander domain and demonstrate empirically that the proposed planning and execution system is capable of commanding a set of realistic surface scenarios as part of a larger Europa Lander surface autonomy software prototype. We expect that an approach to scheduling and execution grounded in decision theory will be an enabling technology for future tightly constrained planetary surface missions.