One of the techniques to fulfill the ever growing demand of high data rates and high spectral efficiency for future wireless networks is to use multiple input multiple output (MIMO) wireless communication systems by employing multiple number of transmitting and receiving antennas. However, conventional MIMO detection techniques like zero forcing (ZF), minimim mean square error (MMSE), minimum mean square error succesive interference cancellation (MMSE-SIC), minimum mean square error ordered succesive interference cancellation (MMSE-OSIC) provide sub-optimal detection and their bit error rate (BER) performance degrades under imperfect channel state information (CSI) at the receiver. Hence, new algorithms and techniques are being developed for robust and non-erroneous detection in MIMO communication systems. In this paper, a robust detection algorithm called modified gravitational search algorithm (MGSA) is proposed for MIMO detection. The proposed MGSA detection algorithm exploits the concept of gravitational search algorithm for MIMO detection. The proposed MGSA technique outperforms conventional detection techniques like ZF, MMSE and MMSE-OSIC under both perfect and imperfect CSI at the receiver. Simulation results validate the performance and robustness of the proposed detection algorithm.