The accuracy of pulse oximetry for the prediction of oxygen saturation of arterial blood in patients with shock has been hardly studied. This study was undertaken to determine if O2 arterial saturation estimated by the Biox 3700 pulse oximeter (SpO2) with an ear probe could reliably substitute for the measurements of O2 saturation (HbO2) with an IL-282 Co-Oximeter in samples of arterial blood obtained from 24 caucasian patients. All patients were mechanically ventilated and needed vasoactive drugs (dopamine and/or dobutamine). Of 24 patients 13 had shock: cardiogenic (n = 6) and septic (n = 7). The mean difference between HbO2 and SpO2 was 2.49% +/- 4.24, with a 95% confidence interval of 0.7% to 4.3% (p = 0.009). There were clinically important differences between both methods since in 9 of 24 patients (37%) SpO2 values were at least 4% lower or greater than HbO2. This disagreement was also apparent in patients with (6/13, 46%) or without shock (3/11, 27%). To conclude, pulse oximetry is not always a sufficiently reliable method to predict HbO2 in patients with or without shock treated with vasoactive drugs.