A feasibility study was carried out to evaluate a chemometrics-enhanced measurement of oxyhemoglobin concentration in whole blood without pre-treatment by lysing cellular components in the sample. Conventional in vitro multi-wavelength CO oximeters pre-process blood by sonication or detergent dilution to lyse blood cells to reduce light scattering. Two limitations result: (1) residual cell membrane fragments can seed surface biofouling and (2) dilution errors can occur. A full wavelength method using multivariate analysis in chemometrics was applied to correct the light scattering effect in the measurement of oxyhemoglobin concentration. Whole blood specimens were adjusted to different oxyhemoglobin concentrations with gas mixtures (N₂, CO₂ and O₂). An Ocean Optics miniaturized spectrophotometer with a 100 µm pathlength optical cell was used for transmission measurements from 500 to 700 nm. Original spectra were smoothed and a second derivative transformation was performed to eliminate the baseline shift and slope changes from light scattering. Indirect calibration was applied to the second derivative spectra. Two-factor cross-validation by principle components regression on two sets of data showed r² = 0.985 and 0.946 between predicted oxyhemoglobin concentration and those measured by an AVL 912 CO oximeter with RSD = 3.85 and 6.83%, respectively. Error analysis gave s = 2.36 × 10^–5 (RSD = 0.23%) on derivative absorbance for the spectrophotometer measurement alone. Specimen settling and specimen sampling gave imprecision on derivative absorbance of s = 6.17 × 10^–4 (RSD = 4.4%) and s = 4.52 × 10^–4 (RSD = 1.4%), respectively.