Earthquake causes great damage to the distribution system, decreasing the quality of power service. Because of geographical characteristics of the outage areas have impacts on the routing of repair crews and logistics support. Therefore, we propose an optimization model for post-earthquake distribution system restoration considering geographical characteristics to maximize the cumulative utility of repairing. A hybrid improved bacterial colony chemotaxis optimization algorithm (HIBCCOA) is designed to solve this model. We take a serial of scenarios derived from IEEE34, IEEE123, and IEEE240-bus system to validate the effectiveness of the model and the performance of the approach by considering different geographical characteristics, different damaged levels, different scales of repair crews. The results show that: 1) The geographical characteristics of the outage areas have impacts on the density of their transportation road network, moreover, the cumulative utility of repairing decreases with the density of their transportation road network decreases; 2) In IEEE34-bus system, HIBCCOA is out performance of genetic algorithm (GA) and the enumerate method (EM) for objective values and CPU times, the rank of them is HIBCCOA, GA, and EM; 3) In IEEE123 and IEEE240-bus system, HIBCCOA is better than GA in the objective values and the CPU times.