For continuous process manufacturing systems (CPMSs) where the production process cannot be stopped, two popular performance evaluation metrics are production efficiency and stability. With the development of sensors and communication technologies, the condition-based decisions can effectively coordinate the operation and maintenance (O&M) management of CPMSs to improve the production completion rate. However, most papers studied condition-based operation (CBO) and condition-based maintenance (CBM) separately which led to the inability to obtain optimal solutions. In addition, the effect of imperfect maintenance on production efficiency and stability has also been ignored. Therefore, this work develops an optimal condition-based operation and maintenance (CBOM) policy for CPMSs. CPMSs are required to complete a series of specified production missions within a finite horizon, and the optional maintenance actions include do nothing, imperfect maintenance, and replacement. The optimization objective to determine the optimal joint O&M policy by maximizing the average production completion rate. In the CBOM policy, the production completion rate of CPMSs under different missions is evaluated by a stochastic flow manufacturing network (SFMN). Since the CPMS has Markov property, we use the Markov decision process (MDP) framework to solve the CBOM optimization problem. The main contributions of this work include: (1) compared with existing studies, a more rational CBOM policy is proposed, which focuses on maximizing the production efficiency and stability of CPMSs; (2) the impact of imperfect maintenance on production efficiency and stability is considered in the CBOM, which makes the proposed policy has better applicability. Finally, the proposed approach is demonstrated in a hot rolling manufacturing system, and a sensitivity analysis of the relevant parameters is also performed.