This article reveals three methods of noise generating with Gauss distribution by using uniformly distributed random-numbers in computer imitation modeling of the digital communications systems – the uniformly distributed random-numbers summing method, Box-Muller method and Marsaglia-Tsang ziggurat-method. The results of simulation by the Monte Carlo include histograms of noise generators and limits for accuracy and computational time of various methods of Gaussian noise generating to evaluate the noise immunity of transmission of binary data by antipodal signals in symmetric channels with intersymbol interference and the Viterbi detection