The Multiple-Input Multiple-Output (MIMO) radar combines the signals from multiple transmit and receive elements separately to form a virtual array. Consequently, system imperfections in the MIMO radar caused by channel imbalance and mutual coupling are repeatedly present in the virtual array, affecting the beamforming performance. In this paper, the beam pattern deviations caused by these imperfections are described mathematically in relation to a single point target. Furthermore, a nonparametric calibration tech-nique to estimate and compensate for these effects at different angles is presented by exploiting that the same imperfections are present multiple times in the virtual array. The proposed calibration method is applied to measurements with an 8 × 8 colocated X-band MIMO radar, and the peak sidelobe level (PSL) and the inte-grated sidelobe level (ISL) are used as performance parameters. The calibration technique is found to improve both the PSL and the ISL by approximately 15 dB within the whole field-of-view (FOV) for antennas that are not minimum scatterers. For angles above ±30∘ from the antenna boresight, the performance of the proposed calibration method is found to surpass other calibration methods. This is advantageous for MIMO radars as they typically cover a wide FOV.