A consideration of neural counting methods.

It is often necessary to obtain unbiased estimates of neuronal or synaptic numbers. In the past, estimates were almost always done by counting profiles of these structures in single histological sections. Assumptions were then made and calculations were done to determine particle numbers or ratios. To the extent that the assumptions deviated from reality, the conclusions will be biased. That these biases are, in fact, serious has recently become apparent. To obtain unbiased particle counts, the presently available methods are serial-section reconstructions (which are accurate but cumbersome), and the recently developed disector method. The disector method, because it is unbiased and easy to use, is becoming the method of choice. The goals of this paper are to show why previous methods are biased and to describe the rationale behind the disector method so that neuroscientists can consider its appropriateness for their work.