This paper presents a methodology to evaluate the influence of downtime on the long-term energy yield of a floating wind farm. It is based on the definition and integration of three models: a discrete event model, which simulates failures and reparation times, a floating wind turbine model, which simulates platform motions and operational stops, and a wind farm model, which models wake effects between turbines. The methodology is applied to a wind farm composed by ten floating turbines and located off the coast of Cantabria, Spain. Results indicate that downtime affects significantly the mean energy yield of the farm, which decreases linearly with the mean failure rate, mean reparation time and probability of exceedance of the operational thresholds for the considered case studies. (C) 2017 Elsevier Ltd. All rights reserved.