In present-day pulse oximeters, oxygen saturation in arterial blood (SpO2) is computed by utilizing an empirical relationship extracted from a calibration curve. The calibration curve is obtained by curve-fitting data acquired from volunteers. A novel method of computation of SpO2 that does not require the use of a calibration curve is presented in this paper. Based on a model for the attenuation of light through skin, tissue, bone, and blood, suitable processing steps are identified so that the analytical expression derived for the estimation of SpO2 becomes free of not only patient but sensor-dependent parameters as well. The experimental results presented in this paper establish the efficacy of the proposed method. © 2009 IEEE.