Rapid development of 3D scanning technologies lead researchers to use them for people recognition, cameras are faster, models are less noisy and with higher resolution. 3D facial models have been widely used for many biometrics applications. Nevertheless 3D face recognition topics mainly assume that scans contain a face and the face is mostly in frontal position. In real world we might have a situation, where a scanned model is not sufficient for recognition. In this article we propose a generic face model validation algorithm which can exclude non-face models from recognition query. The algorithm was tested on more than 1500 range scans including face and non-face models. Obtained results prove, that the generic model validation approach can be used to reject non-face models from the recognition pipeline.