We have explored a new approach for smoothing volumetric seismic dips and azimuths. This method iterates through a loop of three steps: (1) set a guide direction, (2) invert opposite vectors, and (3) average vectors and update the guide. This inverse-vector method can overcome instabilities where the gradient structure tensor approach fails — namely, when a dominant eigenvalue does not exist and the wrong eigenvector is selected to determine orientation. Inverse-vector smoothing is simple to implement and computationally efficient. The resultant dips and azimuths are consistent spatially and thus are more interpretable and suitable for calculating curvature and other dip-based attributes.