Major disparities in reported levels of basal human nitric oxide metabolites have resulted in a recent literature focusing almost exclusively on methods. We chose to analyze triiodide chemiluminescence, drawn by the prospect of identifying why the most commonly employed assay in nitric oxide biology typically yielded lower metabolite values, compared with several other techniques. We found that the sensitivity of triiodide was greatly affected by the auto-capture of nitric oxide by deoxygenated cell-free heme in the reaction chamber. Potential contaminants and signal losses were also associated with standard sample purification procedures and the chemistry involved in nitrite removal. To inhibit heme nitric oxide auto-capture, we added potassium ferricyanide to the triiodide reagent, reasoning this would provide a more complete detection of any liberated nitric oxide. From human venous blood samples, we established nitric oxide levels ranging from 0.000178 to 0.00024 mol nitric oxide/mol hemoglobin. We went on to find significantly elevated nitric oxide levels in venous blood taken from diabetic patients in comparison to healthy controls (p < 0.0001). We concluded that the lack of signals reported of late by several groups using triiodide chemiluminescence for the detection of hemoglobin-bound nitric oxide may not represent levels on the border of assay sensitivity but rather underestimated values because of methodological limitations. We therefore stress the need for assay systems to be developed that differentiate between individual nitric oxide metabolite species and overcome the limitations we outline, allowing accurate conclusions to be drawn regarding physiological nitric oxide metabolite levels.