In recent years, Multiple-Input Multiple-Output (MIMO) systems, which use several transmit and receive antennas, have attracted much attention for high performance radio systems. In MIMO systems, the channel estimation Is Important to distinguish transmit signals from multiple transmit antennas. While the Space-Alternating Generalized Expectation-maximization (SAGE) algorithm Is known to offer good channel estimation and data detection. We proposed earlier the Minimum Mean Square Error (MMSE)-based SAGE algorithm for MIMO systems where the MMSE estimation Is used for channel estimation. We showed that the proposed MMSE-based SAGE algorithm can achieve the better bit error rate (BER) than the maximum likelihood (ML) detection with training symbols. The MMSE channel estimation needs the knowledge of the maximum Doppler frequency Fd for deriving the covarlance matrix of the channel and the variance σ2 of additive white Gausslan noise (AWGN). Additionally, the computation of the MMSE channel estimation requires O(L3) operations where L Is the transmitted frame length. Thus, Its computational complexity Is high. In this paper, we propose a Zero-Forcing (ZF)-based SAGE algorithm for channel estimation and data detection In MIMO systems that does not need the knowledge of Fd and σ2. Since the computation of the proposed ZF-based SAGE algorithm requires O(N3) operations where TV Is the number of transmit antennas, Its computational complexity Is low. We show that the proposed ZF-based tracking SAGE algorithm with less computational complexity can achieve almost the same BER as that of the MMSE-based tracking SAGE algorithm.