High concentrations of glutaraldehyde (2-5%) were found optimal for fixation of glutamate. In the absence of glutaraldehyde, (para)formaldehyde does not permanently retain L-[3H]-glutamate or D-[3H]-aspartate previously taken up into brain slices. Rats were fixed by rapid transcardial perfusion with 2.5% glutaraldehyde/1% (para)formaldehyde, and brain samples osmicated, embedded in epoxy resin, sectioned, and exposed to specific antisera to glutamate (conjugated to carrier protein by glutaraldehyde), followed by colloidal gold-labeled second antibody. The gold particle density was higher over putative glutamatergic nerve terminals than over any other tissue elements (two to three times tissue average in cerebellum and hippocampus). Calibration by test conjugates containing known concentrations of fixed glutamate processed in the same fluid drops as the tissue sections indicated that the concentration of fixed glutamate in putative glutamatergic terminals in hippocampus CA1 was c. 20 mmol/liter. The grain density over the parent cell bodies was only slightly higher than the tissue average. (Grain densities over test conjugates of other amino acids, aldehyde-fixed to brain macromolecules, were similar to that over empty resin. Labeling was blocked by glutamate-glutaraldehyde but not by other glutaraldehyde-treated amino acids.) In other experiments, brain slices were incubated in oxygenated artificial cerebrospinal fluid (CSF) and then immersion-fixed and processed as above. Here, the ration of grain densities in putative glutamatergic terminals vs other tissue elements was greater than in perfusion-fixed material. Comparison of intra-terminal areas poor and rich in synaptic vesicles suggested that in this preparation vesicles contained at least three times the glutamate concentration of cytosol. In the glutamatergic synapses of the giant reticulospinal axons in lamprey the ratio was over 30. Prolonged K+ depolarization of hippocampal and cerebellar slices reduced the nerve terminal glutamate immunoreactivity in a Ca2(+)-dependent manner. The results suggest that glutamate is released by exocytosis at excitatory synapses and show that immunocytochemistry can be used to study the cellular processing of small molecules.