In deeply anesthetized cats, determinations of motor-axonal conduction velocity (CV) were made using extracellular potentials recorded from single, functionally isolated motor axons innervating the muscle tibialis posterior. Axons were activated by suprathreshold electrical stimulation at the ventral-root level. Action potentials were recorded with 3 bipolar electrodes located on the muscle nerve at the level of the popliteal fossa. The most proximal and distal of the bipolar muscle-nerve electrodes were 16.4-22.0 mm apart. Estimates were made of CV from ventral root to muscle nerve (conventional CV) and between the proximal and the distal pairs of muscle-nerve electrodes (muscle-nerve CV). An evaluation was based on comparison of these CVs, estimates of uncertainties in time and distance measurements and simulations of the effects of recording conditions on CV estimates. The analysis indicated that the uncertainty in the conventional CV measurement of mammalian motor axons is at least +/- 2%. However, variability may be as great as 20% between muscle-nerve CV measurements from different experiments, probably due to such factors as regional variation in CV and differences in recording configuration.