Due to the variation of compartment design and occupant’s posture in self-driving cars, there is a new and major challenge for occupant protection. In particular, the studies on occupant restraint systems used in the self-driving car have been significantly delayed compared to the development of the autonomous technologies. In this paper, a numerical study was conducted to investigate the effectiveness of three typical restraint systems on the driver protection in three different scenarios. It is found that based on the simulation results: (1) All the restraint systems are capable of providing effective protection for the driving driver and the 4-point belt restraint system has advantages due to its better protective effect on the occupant thorax; (2) When the driver is in half-reclining and reclining resting modes, head HIC36, neck Nij and chest compression are about 572.9-1524.3, 0.64-1.47, and 14.7-48.3 mm, respectively; These values are higher than those of a driving driver by 0.2%-198.3% for HIC36, 113.3%-359.4% for neck Nij, and -59.6%-79.8% for chest compression, respectively. (3) There is an evident “submarining” of dummy in the half-reclining and reclining resting modes. This study shows that the mainstream restraint system cannot meet the safety requirements for occupants with multi-postures in the self-driving cars. Smart restraint systems which can be actively adjusted for protecting occupants sitting in different postures are deemed to be necessary in case of unavoidable crashes.