In multiple-inputmultiple-output (MIMO) communications, the notion of asymmetric channel refers to the situation when the number of transmit antennas is strictly larger than the number of receive antennas. Such channels can often be found in MIMO downlink transmissions. While existing cyclic-division-algebra (CDA)-based codes can still be employed to achieve the optimal diversitymultiplexing tradeoff (DMT) at high signal-to-noise ratio (SNR) regime, such codes cannot be directly decoded using, for example, the pure sphere decoding method. Although other means of decoding methods such as minimum mean square error generalized decision feedback equalizer (MMSE-GDFE) with lattice search and regularized lattice decoding are available, an alternative approach is to constrain the number of active transmit antennas in each channel use to be no larger than the number of receive antennas. The resulting system is coined constrained asymmetric MIMO system. Two general types of asymmetrical channels are considered in this paper, namely, 1) when there are two receive antennas and the number of transmit antennas is arbitrary, and 2) when the number of transmit antennas is one larger than the number of receive antennas. Explicit optimal transmission schemes as well as the corresponding code constructions for such constrained asymmetric MIMO channels are presented, and are shown to achieve the same DMT performance as their unconstrained counterparts.