Recently several methods have been proposed for the system identification of wind turbines which can be considered as a linear time-varying system due to the operating conditions. For the identification of linear wind turbine models, either black-box or grey-box identification can be used. In the case of black-box identification no "physical" model structure is assumed and a large number of parameters have to be defined by the identification techniques. For the identified black-box models, it is difficult to perform model analysis since the internal states are not known to the user. The operational model analysis (OMA) methodology can provide accurate estimates of the natural frequencies, damping ratios and mode shapes of the systems as long as the measurements have a low noise to signal ratio. However, in order to take information about the wind turbine into account a grey-box identification method can be used. This paper reviews proposed system identification methods and the durability and the limitations of the different methods are outlined.