NADH oxidase activity has been detected at the ultrastructural level using cerium ions to trap H2O2 generated by the enzyme (via intermediate reactive oxygen species). In an attempt to localize NADH oxidase activity at the light microscope level using the cerium-diaminobenzidine (DAB)-nickel-H2O2, the cerium-DAB-cobalt-H2O2 or the cerium-alkaline lead procedures, the distribution patterns of the revealed enzyme were found to be identical to those for non-specific alkaline phosphatase and especially 5'-nucleotidase activity. With the cerium-DAB-cobalt-H2O2 visualization procedure, the distribution pattern of the final reaction product was similar to that obtained with the other two techniques but much less final reaction product was formed. Incubations for NADH oxidase activity performed in the presence of exogenous catalase or in the absence of catalase or peroxidase inhibitors did not affect the staining intensity, whereas inhibitors of 5'-nucleotidase (EDTA) and non-specific alkaline phosphatase (levamisole) always did. Therefore, phosphatases contribute to the formation of the final reaction product. Since NADH initially cannot be hydrolysed by either of these two phosphatases, then presumably nucleotide pyrophosphatase (E.C.3.6.1.9) cleaves NADH into 5'-AMP and nicotinamide mononucleotide in a first step. Both nucleotides can be hydrolysed further by the two monophosphatases. These then generate cerium phosphate which is detected by the DAB-nickel-H2O2, DAB-cobalt-H2O2 or lead visualization methods.