Speech perception degrades in background noise, which is primarily caused by an inability to identify the consonants. In this study, we aim to determine how different consonants are represented in the spiking patterns of auditory nerve fibers and how these representations are affected when presented in speech-shaped background noise.Single-unit auditory nerve fiber recordings were collected from fourteen normal-hearing young-adult gerbils, using glass electrodes. Three different naturally-spoken vowel-consonant-vowel constructs were presented in quiet and in background noise at 5 dB SNR, imitating a realistic communication situation. Previous behavioral studies in gerbils using the same stimuli revealed that discrimination between consonants was significantly compromised at this SNR.The nasal consonant /m/ was represented by a rate and a temporal code that differed significantly from the rate and temporal profiles of responses to /b/ and /t/. Background noise eliminated these rate differences, while the temporal code during /m/ remained significantly different compared to /b/ and /t/. Neurograms revealed a sharp peak in discharge rate across a broad frequency range for the plosive consonants /b/ and /t/. This, however disappeared with background noise due to rate saturation.In summary, the consonants were encoded differently by the spiking patterns of the auditory nerve, and background noise thus also degraded these codes differentially. Our results suggest that the behavioral difficulties to discriminate these consonants in background noise originate peripherally.