In contrast with results obtained in experimental animals, antibodies to microtubule associated protein-2 (MAP2) preferentially label abnormal structures in human nervous system tissue samples, but the normal sites at which MAP2 is expressed are not well-defined. To determine the distribution of MAP2 in the human central (CNS) and peripheral (PNS) nervous systems, we prepared monoclonal antibodies (MAbs) specific to MAP2, and compared the localization of this MAP in postmortem bovine and human tissues as well as in several human neural cell lines that express either neurofilament (NF) or glial filament (GF) proteins. Eight MAbs specific for phosphate-independent epitopes in bovine and human MAP2 were obtained, and those that performed well in tissues produced immunoreactivity confined to the somatodendritic domain of neurons in bovine and human CNS and PNS tissues. Other neural cells (e.g. astrocytes) did not express MAP2 immunoreactivity using these MAbs. Postmortem delays of less than 24 h prior to tissue denaturation did not affect the distribution of MAP2 immunoreactivity. However, microwave denaturation of these tissues preserved MAP2 immunoreactivity better than fixation with Bouin's solution or formalin. Microwave treatment also improved the immunoreactivity of several MAbs for NF and GF proteins. Finally, MAP2 was not detected in human neural cell lines that express NF (2) or GF (1) proteins. We conclude that microwave denaturation provides an effective means to preserve the immunoreactivity of normal human neuronal cytoskeletal proteins, and that this method of tissue denaturation allows the normal distribution of MAP2 to be defined in postmortem samples of human CNS and PNS tissues.