An estimator of the total number of synapses in neocortex of human autopsy brains based on unbiased stereological principles is described. Each randomly chosen cerebral hemisphere was stratified into the four major neocortical regions. Uniform sampling with a varying sampling fraction in each region of neocortex was performed. The total volume of each neocortical region was estimated using point counting according to Cavalieri's principle. The ethanolic phosphotungstic acid staining technique was modified for synapses in human autopsy brains. The numerical density of synapses in each neocortical region studied was estimated using the disector at the electron microscopical level. The total number of neocortical synapses in each region was estimated as the product of the total volume of neocortex and the numerical density of synapses. The influence of the postmortem fixation delay on the number of synapses was investigated in five large mammals (one dog, one cow, and three pigs), the brains of which were kept under conditions similar to those under which human corpses are normally kept. The apparent decrease of 3.9% in the numerical density of synapses in the large mammals following a 2‐day fixation delay was not significant. The average total number of synapses in the neocortex of five young male brains was 164 × 10¹² (CV = 0.17). An analysis of the precision of the estimate of the total number of synapses in neocortex indicates that blocks represent both the major source of variation and the largest workload. Using eight blocks per brain the imprecision of the estimate is, however, only 66% of the total variance. Synapse 41:258–273, 2001. © 2001 Wiley‐Liss, Inc.