Previous theoretical and empirical studies suggest that CEOs' political connections are valuable to firms. We examine whether such connections become entrenched if the expected political capital fails to materialize and the firm lacks other types of political power. Using a sample of listed non-SOEs in China, we show that politically connected CEOs have a lower probability of turnover and cause a weaker turnover-performance sensitivity than non-politically connected CEOs. Further analyses show that these turnover patterns are not consistent with alternative explanations, such as superior managerial ability, being a member of controlling families or being promoted from the inside. The turnover patterns are less pronounced in firms with alternative political power, such as connected boards or being vital to the local economy. Following the turnover of politically connected CEOs, firm performance does not necessarily undergo significant improvement. Our results call for new theories that comprehend the real effects of political connections.