A Gaussian regression algorithm is proposed for reducing bit error rate (BER) measurement time and facilitating, by extrapolation, a bit error probability estimation as a function of the signal/noise ratio. The application of this algorithm is illustrated in two BER measurement methods using noise injection or decision threshold level shifting. The overall performance assessment of a system (error probability as a function of the signal/noise ratio and noise standard deviation in the absence and in the presence of the signal) are thus obtained.