Compared to conventional marine systems, where the onboard crew can perform frequent and flexible maintenance, autonomous marine systems (AMS) only involve a limited (or, even no) crew during a voyage, and this challenges maintenance planning and execution. The current study identifies the relevant issues and proposes to solve these through developing a dynamic maintenance planning method for AMS. By considering economical dependencies among components, the study presents a dynamic grouping method to determine the optimum maintenance opportunities for AMS in the future. Stochastic dependencies of components are considered by using the Markov model. A multiphase Markov model is proposed for modeling stochastic dependencies between components where the limited and irregular maintenance opportunities are handled by the multiphase part of the model. A heuristic method is proposed to deal with the combinatorial challenge.To demonstrate the application of the proposed method, the maintenance planning of a cooling system of an autonomous ship is performed in a case study. To validate its performance, the proposed heuristic method is compared with existing ‘short-sighted’ methods for a selection of candidate groups for maintenance. In the validation, various scenarios with different component states and maintenance strategies are tested.