We study probabilistic recursive feasibility guarantees for model predictive control under uncertain environments, with applications to autonomous driving.