The research objective is to study the correcting capacity of a new soft decoder of Reed-Muller codes. The technique of achieving the goal is an experimental study of the decoder using a specially built simulation model of the antinoise data transfer channel. The model message source and receiver handle the numeric data identified in F3 field. The communication line of the constructed model produces digital or continuous signals depending on the settings. In the case of continuous signals, two variants of the basic signal distortion and their combinations are considered. Noise immunity of the simulated communication link is provided by using Reed-Muller second-order codes identified over F3 field, and the new soft-decision decoder for these codes. The results of the simulation experiments show that the decoder under study in both digital and semicontinuous channel allows correcting more errors than it is guaranteed by the minimum code distance. The decoder has proved the most effective in the semicontinuous communication channels. The decoder’s correcting capacity depends heavily on the communication line type and on the signal distortion mode; it is not sensitive to the error location within the codeword. Conclusions are made on the use of the new decoder in the low-rated communication channels to provide noise immunity, and in a number of the cryptographic applications.