The paper presents a novel method for statistical quantification of wind power variation based on a new concept of homogeneous generation periods (HGPs). It starts with the state-of-the-art review of approaches addressing the quantification of wind power variation. The suggested method is based on identification of (a) wind generation gradients (positive, negative, insignificant and ineligible) for the time steps and (b) HGPs (increasing, decreasing, constant and null-generation). In result, the method provides a statistical evaluation on variation of wind generation process in observation period in terms of statistical characteristics of HGPs. The validation of method is performed on a numerical set-up representing Lithuanian wind power cluster in a windy 13-day period. The validation results point to the effectiveness of the suggested method for the ex-post quantification of wind generation variation in a short-term time span. (C) 2014 Elsevier Ltd. All rights reserved.