A detailed numerical analysis of the Edge Crack Torsion test was performed to verify its adequacy for the mode III interlaminar fracture characterization of composite laminates. A new data reduction scheme based on specimen compliance was proposed. Three-dimensional finite element analyses including a cohesive damage model were performed in order to evaluate the accuracy of perceived G(IIIc) values obtained with the proposed method. Despite some degree of crack length dependency of perceived G(IIIc) acceptable errors could be achieved within a certain crack length range, which is thus recommended for experimental tests.