Hepatopulmonary syndrome, arterial hypoxemia caused by intrapulmonary vasodilatation, occurs in approximately 10% of patients with cirrhosis. The severity of hypoxemia affects liver transplant candidacy and is associated with increased morbidity and mortality posttransplantation. Screening guidelines for detecting the presence of arterial hypoxemia do not exist. The aim of this study is to investigate the accuracy and utility of pulse oximetry in the detection of hypoxemia (PaO(2) < 70 mm Hg) in patients with cirrhosis. Two hundred prospective liver transplant candidates were compared with 94 controls. Arterial oxyhemoglobin saturation was obtained by pulse oximetry (SpO(2)) and compared with simultaneous arterial blood gas (ABG) oxyhemoglobin values (SaO(2); bias = the difference). PaO(2), carboxyhemoglobin, methemoglobin, and routine clinical and biochemical parameters were investigated to account for the bias. SpO(2) overestimated SaO(2) in 98% of patients with cirrhosis (mean bias, 3.37%; range, -1% to 10%). Forty-four percent of patients with cirrhosis and controls had a bias of 4% or greater. No clinical or biochemical parameters of cirrhosis accounted for the overestimation of pulse oximetry. Twenty-five subjects with cirrhosis were hypoxemic, and an SpO(2) of 97% or less showed a sensitivity of 96% and a positive likelihood ratio of 3.9 for detecting hypoxemia. An SpO(2) of 94% or less detected all subjects with an arterial PaO(2) less than 60 mm Hg. Pulse oximetry significantly overestimates arterial oxygenation, and the inaccuracy is not influenced by liver disease. Nevertheless, pulse oximetry can be a useful screening tool to detect arterial hypoxemia in patients with cirrhosis, but a higher threshold for obtaining an ABG must be used.