Numbers of primary afferent neurons underlie important generalizations concerning the organization of primary sensory systems. A major difficulty, however, is that different investigators do not agree on the neuronal counts. The problem, in our opinion, is that the various methods used to determine these numbers do not provide the same results. Thus to be certain that a method provides accurate counts, calibration is necessary. To do this, true numbers of ganglion cells were determined by serially reconstructing significant parts of four rat lumbar dorsal root ganglion cell populations. Then 6 commonly used methods of counting neurons were used to determine neuron numbers for these same populations. The data indicate that the empirical method, using the modifications recommended in this paper, estimates numbers of neurons with the needed accuracy whereas the other 5 do not. Thus, of the tested counting procedures, the empirical method is recommended. If other methods are to be used, they should also be calibrated.