This study investigates the optimization of the operation and maintenance of offshore wind turbines based on condition monitoring data. Due to their increasingly remote and challenging location, a decision framework is proposed that optimizes the cost and risk of maintenance scheduling based on, dynamic Bayesian network based, iterative estimation of turbine lifetime. This allows for the combining of predictive and opportunistic maintenance strategies, scheduling preventative component replacements to minimize lost production, while maximizing lifetime and optimizing use of resources. Assessment of related literature and applications suggests the approach could lead to a reduction of maintenance costs that exceeds 30%. The proposed framework relies on effective fault detection and prognosis of wind turbine components, realised through the implementation of machine learning techniques on the turbine’s own SCADA system. The installing of additional sensors can potentially increase the capability of this system for more advanced diagnosis and localization of a fault.