This article proposes a methodology for evaluating the reliability of offshore wind farms (OWF) connected to oil and gas platforms, to provide a more sustainable and cleaner production by producing part of the energy needed for supplying the internal services through renewable energy. Different collector network topologies are compared in order to calculate the availability of supply at the Point of Common Coupling.The evaluation is carried out by combining Markov processes to capture wind speed uncertainty with the analytical theory of the Universal Generating Function and calculates the Generation Ratio Availability and Energy Not Supplied (ENS) indices.The design of an 8 MW OWF located in northeastern Brazil is used as a test, considering two correlated wind speeds and an equivalent single wind speed. The results show that the single-sided ring collector network topology is the most reliable, while the radial one is the least reliable. The results of a sensitivity analysis of each topology to the unavailability of the collector network and the wind turbine are also presented. The use of actual correlated wind speeds instead of the equivalent one results in higher ENS and, therefore, a less reliable system. A cost analysis shows that the star topology has the highest levelized cost of electricity.