Most studies consider voltage-gated ion channels biophysics as deterministic. However, individual ion channels undergo thermodynamic fluctuations which provoke stochastic transitions between closed and open states. This “channel-noise” leads to fluctuations of the membrane potential (Vm) and to unpredictability of the neuronal output. Channel noise has been mostly studied through somatic recordings which may underestimate it due to the great somatic capacitive load and the large somatic channels amount leading to an averaging of the individual channel stochasticity. Computational studies have suggested that thinner compartments such as axons are highly sensitive to channel noise but experimental evidence is lacking. By performing dual soma-axon wholecell recordings in cortical layer 5 pyramidal neurons, we observed larger subthreshold Vm fluctuations in the axon than in the soma. Fluctuations increased with voltage depolarization and decreased following TTX application, strongly suggesting that they were due to channel noise. Using a simple computational model we showed that the larger axonal Vm fluctuations were due to the small axonal capacitive load. Unexpectedely increasing the ion channels amount lead to an increase of axonal Vm fluctuations. Therefore, we showed that channel noise has a bigger impact on axonal than somatic Vm which may be important for both spike initiation and spike propagation dynamics.