Abstract We propose in this paper a new objective metric for the visual quality assessment of 3D meshes. The metric can predict the extent of the visual difference between a reference mesh, which is considered to be of perfect quality, and a distorted version. The proposed metric is based on a mesh local roughness measure derived from Gaussian curvature. The perceptual distance between two meshes is computed as the difference between the normalized surface integrals of the local roughness measure. Experimental results from three subjective databases and comparisons with the state of the art demonstrate the efficacy of the proposed metric in terms of the execution time and the correlation with subjective scores. Finally, we show a simple application of the metric in which it is used to automatically determine the optimum quantization level of mesh vertex coordinates.