The increasing penetration of renewable energy sources (RESs) is significantly impacting the performance of traditional control and protection schemes employed in power systems. The early detection and size estimation of loss of generation (LoG) events contribute to the timely provision of remedial actions required to preserve the frequency stability of the system. This paper proposes an effective method for locating and estimating the size of LoG events in power systems with high penetration of RESs. A system of linear equations for every candidate LoG location is formulated based upon KVL and KCL equations. Suitable nodal current sources are employed to model RESs and the candidate tripped generator in the superimposed circuit. The nodal current injections of RESs are estimated based on their power setpoints and available measurements provided by PMUs. The solution of the systems of equations established provides the LoG location and size using the least-squares method. The proposed method is more accurate than existing LoG identification methods that resort to frequency measurements and the knowledge of system inertia. Centralized under-frequency load shedding is presented as a potential application for the proposed method. Extensive simulations conducted on the IEEE 39-bus test system verify the effectiveness and accuracy of the proposed method.