The effective life management of large and diverse fleets of wind turbines is a new problem facing power system utilities. More specifically, the minimization of over-speed risk is of high importance due to the related impacts of possible loss of life and economic implications of over-speed, such as a loss of containment event. Meeting the goal of risk minimization is complicated by the large range of turbine types present in a typical fleet. These turbines may have different pitch systems, over-speed detection systems, and also different levels of functional redundancy, implying different levels of risk. The purpose of this work is to carry out a quantitative comparison of over-speed risk in different turbine configurations, using a Markov process to model detection of faults and repair actions. In the medium-long term, the risk associated with different assets can be used as a decision-making aid. For example, if the operator is a utility, it may want to avoid purchasing high-risk sites in the future, or may need to develop mitigation strategies for turbines at high risk of over-speed.