Using current Medicare guidelines for the prescription of long-term oxygen therapy, we studied the impact on decision-making of substituting cutaneous oxyhemoglobin saturation measurements (SaO2) for direct arterial oxygen tension measurements (PaO2). Fifty-five patients with chronic lung disease and resting hypoxemia were studied. More than 80 percent of patients with a resting PaO2 of 7.33 kPa (55 mm Hg) or less had a cutaneous oximetry SaO2 greater than 85 percent. These patients would not have met the guidelines for long-term oxygen therapy if the cutaneous oximetry measurements were used instead of direct PaO2 measurements. Substituting a threshold criterion of 88 percent instead of 85 percent resulted in fewer patients being denied oxygen therapy but also included patients with PaO2 values greater than 7.33 kPa (55 mm Hg). We conclude that cutaneous oximetry cannot be substituted equivalently for PaO2 measurements in prescribing long-term oxygen therapy.