In this thesis, we examine the performance of convolutional coded DS/SSMA communication system with error-and-erasure decoding. The performance evaluation is carried out over additive white Gaussian noise channel, multipath Rayleigh fading channel with and without signal fading amplitude information. The channel state information (CSI) considered are the demodulator output signal strength and the signal fading amplitude. The demodulator makes a three-level-decision {-1,1,?} based on the CSI, and the optimum decision scheme(threshold) that minimizes BER is found in terms of channel traffic. A simple parallel decision structure is proposed to give a performance which is very close to the optimum decision scheme. It is shown that the decision scheme that maximizes the channel cut-off rate $R_o$ is almost optimal. For fading channels, the erasure decision scheme based upon the estimate of the demodulator output signal strength gives a better performance than the erasure decision scheme based upon the fading amplitude estimate.