Reproducible visualization of neurons and glia in human brain is essential for quantitative studies of the cellular changes in neurological disease. However, immunohistochemistry in human brain specimens is often compromised because of prolonged fixation. To select cell lineage-specific antibodies for quantitative studies of neurons and the major types of glia, we used 29 different antibodies, different epitope retrieval methods, and different detection systems to stain tissue arrays of formalin-fixed human brain. The screening pointed at CD45/leukocyte common antigen (LCA), CD68(KP1), 2',3' cyclic nucleotide phosphatase (CNPase), glial fibrillary acidic protein (GFAP), HLA-DR, Ki67, neuronal nuclei (NeuN), p25alpha-antigen, and S100beta as candidates for future cell counting purposes, because these markers visualized specific neuronal and glial cell bodies. However, significant negative correlation between staining result and formalin fixation was observed by blinded scoring of staining for CD45/LCA, CNPase, GFAP, and NeuN in brain specimens fixed by immersion and stored up to 10 years in 4% formalin solution at room temperature, independent of donor sex and postmortem interval. In contrast, improved preservation of NeuN and CNPase staining, and full preservation of GFAP and CD45/LCA staining in tissue fixed by perfusion and stored for up to 3 years in 0.1% paraformaldehyde solution at 4C, indicated that immunohistochemistry can be performed in well-preserved biobank material.