A standard approach to assessing nitric oxide synthase (NOS) activity in tissue homogenates is 1) removal of small-molecular-weight substances by size-exclusion chromatography, 2) adding back of substrates/cofactors in precise concentrations with a radioactive isotope of arginine (Arg), and 3) quantification of labeled citrulline (Cit) after separation of Arg and Cit by cation-exchange column chromatography. Using this approach and L-[2,3-3H]Arg, we found that the major product(s) was not Cit in cortical homogenates prepared from rat, mouse, and human kidneys. The product(s) mimicked Cit, insofar as it passed freely through cation-exchange columns and comigrated with Cit on both one-dimensional and two-dimensional straight-phase thin-layer chromatography systems. However, it was clearly resolved from Cit by precolumn derivatization and reverse-phase HPLC. The maximum velocity and Michaelis-Menten constant were approximately 100 pmol x mg protein(-1) x min(-1) and 100 microM, respectively, in renal cortical homogenates from rats. The enzyme activity was the same in the presence or absence of cofactors including Ca2+, calmodulin, tetrahydrobiopterin, and NADPH. It was only modestly inhibited by L-Arg analogs and was mainly in the supernatant after a 100,000 g centrifugation. These enzyme characteristics contrasted markedly with those simultaneously obtained for NOS activity in placental homogenates. Thus results from the conventional NOS activity assay should be viewed cautiously.