We present an automated method of spatially detailed 3D asymmetry quantification of face surfaces obtained in a stereophotogrammetric system, and the method was applied to a population of children with juvenile idiopathic arthritis (JIA) who have involvement of one temporomandibular joint (TMJ). Non-rigid registration of a patient face surface to a mirrored version of the same face surface provided a spatially dense map of left-right point correspondences that were used in the asymmetry calculation. For inter-patient comparison and modeling, detailed point correspondence was obtained by non-rigid registration of a symmetric atlas surface to a symmetric version of the patient surface. The method was validated by comparison to a landmark based method of asymmetry quantification, and the two methods were seen to provide similar results. The method was seen to be able to quantify typical patterns of asymmetry in JIA subjects with affected TMJ and compare them to a control population.