In human brain, antibodies to tau proteins primarily label abnormal rather than normal structures. This might reflect altered immunoreactivity owing to post-mortem proteolysis, disease, or species differences. We addressed this issue by comparing the distribution of tau in bovine and human post-mortem nervous system tissues and in human neural cell lines, using new monoclonal antibodies (MAb) specific for phosphate-independent epitopes in bovine and human tau. In neocortex, hippocampus, and cerebellum, immunoreactive tau was widely expressed but segregated into the axon-neuropil domain of neurons. In spinal cord and peripheral nervous system, tau immunoreactivity was similarly segregated but less abundant. No immunoreactive tau was detected with our MAb in glial cells or in human neural cell lines that express neurofilament or glial filament proteins. Post-mortem delays in tissue denaturation of less than 24 hr did not affect the distribution of tau, but the method used to denature tissues did, i.e., microwave treatment preserved tau immunoreactivity more effectively than chemical fixatives such as Bouin's solution, and formalin-fixed tissue samples reacted poorly with our anti-tau MAb. We conclude that the distribution of tau proteins in human nervous system is similar to that described in perfusion-fixed experimental animals, and that visualization of normal immunoreactive tau in human tissues is critically dependent on the procedures used to denature post-mortem tissue samples. Furthermore, microenvironmental factors in different neuroanatomical sites may affect the regional expression of tau.