Objective. Pulse dye densitometry (PDD) uses two wavelengths (805 and 890 nm) in association with pulse oximetry to compute the arterial blood concentration ratio of indocyanine green (ICG) to hemoglobin (Hb). When Hb is measured in the usual way, this permits the PDD to compute cardiac output, plasma or blood volume, and liver blood flow following an intravenous injection of ICG. In this study, we evaluate the accuracy of the PDD calculation of dye concentration by comparing it with measurement of the dye concentration in blood (Cb) measured by the spectrophotometric cuvette method during dye clearance in patients. Methods. In 25 patients receiving major abdominal surgery, ICG (10, 20, or 40 mg) was injected into a central vein and arterial ICG concentration was continuously and simultaneously monitored at nose and finger by PDD; concurrently, ICG concentrations were measured by a spectrophotometer at 805 nm in 4 radial arterial blood samples. Repeated measures or oneway ANOVA were used for comparison of ICG concentrations and percent errors by PDD at the nose, finger, and Cb. Results. The percent error (bias) of calculated dye concentration and its standard deviation (precision) was –3.9 ± 16.8% (p< 0.01)="" with="" the="" probe="" on="" a="" nostril="" and="" 3.4="" ±="" 12.6%="" using="" the="" finger="" probe.="" these="" errors="" were="" found="" to="" be="" greatest="" when="" the="" mean="" transit="" time="" of="" the="" dye="" was="" rapid="" (–20.7="" ±="" 6.8%="" at="" nose="" p="">< 0.01="" and="" –8.5="" ±="" 2.5%="" at="" finger="" p="">< 0.05)="" due="" to="" factors="" other="" than="" the="" time="" delay="" of="" blood="" sampling.="" conclusion.="" these="" errors="" are="" of="" similar="" size="" to="" those="" associated="" with="" thermal="" cardiac="" output="" measurement,="" suggesting="" that="" pdd="" should="" be="" valuable="" clinically="" as="" a="" noninvasive="" tool="" especially="" since="" it="" provides="" values="" for="" blood="" volume="" and="" liver="" blood="">