In this paper, we study the problem of parametric channel estimation for multi-user 3D millimeter wave (mmWave) massive multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems. Modeling the channel by a finite number of multipath rays, we propose a new method to jointly estimate the elevation-azimuth angle and path delay parameters for a desired user. In the proposed method, a new version of subcarrier smoothing (SCS) is developed to construct a full rank correlation matrix. Then the 3D parameters are estimated by using the ESPRIT algorithm. Finally, simulation results are finally presented to verify the efficacy of the proposed algorithm.