Of eight methods examined for measuring plasma hemoglobin in micromolar concentration, all exhibited acceptable linearity, reproducibility, and concurrence except when specimens were icteric or lipemic or contained methemoglobin or methemalbumin. Measurement of absorbance at 578 nm with an Allen correction permits precise assay of plasma oxyhemoglobin concentration as low as 0.01 g/L (1 mg/dL, 0.16 mumol/L), unaffected by hyperlipidemia or hyperbilirubinemia. Discrepancies between methods occurred in 11.6% of a consecutive series of 50 nonicteric patients' plasma specimens. Examination of absorption spectra is helpful when discrepancies are observed between methods. The presence of methemalbumin or methemoglobin in plasma is not recognized by methods that measure only oxyhemoglobin. Increased ceruloplasmin or beta-carotene does not significantly affect results.