The objective is to design efficient coded modulation systems with sub-optimum but practical decoding schemes for Rayleigh fading channels. The well-known rate design rule is conventionally used for additive white Gaussian noise (AWGN) channels. In this paper its application is extended to the Rayleigh fading case by modelling the channel as a time-invariant additive non-Gaussian noise channel, and the equivalent channel capacity equations are evaluated. The sub-optimumbut practical decoding schemes of multistage decoding and parallel decoding on levels are considered. The fading channel is modelled as a time-invariant additive non-Gaussian noise channel by assuming that the channel state information is available at the receiver. The equivalent channel capacity equations for both decoding schemes and a code design example are presented. It is demonstrated that very good coding gains can be achieved if code rates are selected correctly.