The spiral ganglia of the cat, gerbil, mouse, rat, and human were immunohistochemically stained with various monoclonal neurofilament antibodies. Three antibodies to the 200-kD neurofilament protein (R-3, Dräger et al., '84; ICN anti-200, clone NE14, Debus et al., '83; RT-97, Wood and Anderton, '81) labeled the somata of type II spiral ganglion neurons but not those of type I ganglion neurons. In the extreme base of the cochlea of cats, mice and rats, there was intense labeling of a few (less than 0.5% of the total ganglion population) large neurons resembling type I ganglion neurons. Several other neurofilament antibodies (Amersham anti-68, Amersham and ICN anti-160, and SMI-32) did not specifically label type II ganglion neurons but instead labeled all neurons of the spiral ganglion. These two patterns of labeling prompted us to investigate the cause for this difference. Because antibodies against the 200-kD neurofilament protein preferentially labeled type II neurons and because 200-kD neurofilament is highly phosphorylated, we treated cochlear tissue with alkaline phosphatase in order to remove phosphate groups. This treatment eliminated the intense labeling of type II neurons with R-3, ICN anti-200, and RT-97, but had no effect on the intense labeling of ganglion cell bodies observed with the other neurofilament antibodies tested. This evidence suggests that labeling occurs because of the cytoplasmic presence of phosphorylated 200-kD neurofilament protein in type II ganglion neurons. Populations of neurons may thus differ in their neurofilament epitopes and monoclonal antibodies can be used to mark such differences.