Venous oximetry catheters provide useful realtime information about mixed venous hemoglobin saturation (Svo2). Currently available systems utilize either two or three wavelengths of light to obtain these measurements. Previous animal and clinical studies have attempted to compare the accuracy of these two devices under similar circumstances. However, the relative accuracy of the two-wavelength versus three-wavelength systems has never been assessed under identical conditions. For this purpose, we designed an animal model for simultaneous measurement of Svo2, over a wide range of physiologic and pathologic states. Seven anesthetized swine underwent simultaneous placement of two- and three-wavelength catheters. Paired data points consisted of values obtained from a reference oximeter and from each of the catheters. Observations were obtained every 15 min during the following manipulations: (1) eucarbic hypoxia induced by reducing FiO2 to 0.18, 0.15, and 0.12 for 15 min each; (2) stimulated surgical manipulation; and (3) hypovolemic shock produced by hemorrhage to a mean arterial pressure of 50 torr for 1 hr. Data were analyzed by calculation of mean error (bias) and precision for each system in comparison with the oximeter. The overall error of the two-wavelength system was +0.15%, with a precision of +/- 2.52%. The three-wavelength system had an overall error of +3.71%, with a precision of +/- 2.30%. Overall correlation between catheter Svo2 and oximeter values was the same for both devices (r = 0.99). Both currently available in vivo spectrophotometric systems are capable of producing satisfactory results over wide ranges of Svo2. In contradistinction to older reports, we found that the two-wavelength Svo2 system produced results equivalent to those obtained from the three-wavelength device. In this regard, there is no detectable advantage in accuracy to measuring in vivo Svo2 with three rather than with two wavelengths.