In many situations, multiple-input-multiple-output with orthogonal frequency division multiplexing (MIMO-OFDM) channels tend to be spatially correlated due, for example, to limited scattering. Prior knowledge of this channel spatial correlation and the channel frequency correlation can be exploited by using the linear minimum-mean-square-error (LMMSE) technique. However, the complexity of the 2-D LMMSE technique, which fully utilizes both the channel spatial and frequency correlation is quite high. To solve this problem, this paper presents and analyzes several low-complexity, suboptimal, approximated LMMSE channel estimation techniques in the angle domain, where the channel model lends itself to a physical interpretation. The choice of angle-domain techniques is largely dependent on the extent of channel stochastic information (e.g., channel correlation or power) that is available to the receiver. Nevertheless, all the proposed angle-domain techniques have much lower complexity compared to the 2-D LMMSE technique. Further, all the angle-domain techniques improve over the conventional least square (LS) technique for all the typical MIMO-OFDM models under consideration. More importantly, our simulation results show that the angle-domain quasi 1-D (Ql-D) LMMSE technique can achieve similar performance compared to the 2-D LMMSE technique for all typical MIMO-OFDM models with significantly lower complexity.