Abstract— In previous work, we studied the structure and performance of optimum maximum-likelihood receivers for binary antipodal and orthogonal signals in the presence of Gaussiandistributed channel estimation error and additive white Gaussian noise for flat Rayleigh fading channels. In this paper, we investigate the structure and performance of these receivers for an application where Gaussian-distributed channel estimation error arises: minimum mean-square error channel estimation in quasistatic Rayleigh fading channels. Exact closed-form analytical expressions are derived for the average bit error probability (BEP). We quantify the impact of number of pilot symbols in each frame as well as the ratio of the power of the pilot symbol to the power of data on the average BEP. We derive conditions under which orthogonal signalling results in a lower average BEP compared with binary antipodal signalling.