Though peripheral conduction velocity is widely used to characterize afferent fibers according to somatosensory modality, disagreement exists as to whether or not conduction velocity varies along such an axon's length. Therefore, in this experiment, conduction velocities were measured over very short axonal segments (7.5 to 15 mm) within the posterior tibial nerve, sciatic nerve, and L7 dorsal root, using the method of spike-triggered averaging of neurograms recorded from tripolar electrodes. The conduction velocities for several units were also determined using electrical stimulation, so that the accuracy of the two techniques could be compared. For most units, dorsal root and sciatic nerve conduction velocities were not significantly different; however, they were not tightly correlated. Tibial nerve conduction velocity averaged 86% of that within the sciatic nerve. Variations in sciatic nerve conduction velocity within adjacent axonal segments (9 mm in length) rarely exceeded experimental error. It appears that spike-triggered averaging of signals from tripolar electrodes separated 15 mm or more apart is a more accurate method for measuring conduction velocity than electrical stimulation, which was subject to several large errors.