BACKGROUND AND OBJECTIVE:Oxyhaemoglobin saturation of arterial blood is commonly measured using a finger sensor attached to a pulse oximeter (SpO(2)). We sought to compare SpO(2) measured using finger and forehead sensors with oxyhaemoglobin saturation in arterialized capillary samples (ACS) in people with chronic obstructive pulmonary disease (COPD) during exercise. METHODS:During aerobic exercise, SpO(2) was measured continuously by two pulse oximeters: one connected to a finger sensor and to a forehead sensor. Before and after the task, ACS were collected to provide a minimally invasive reference measure of oxyhaemoglobin saturation. Patients with COPD were eligible for inclusion if they desaturated when walking by >4% from resting levels to <90%. Current smokers and those prescribed supplemental oxygen were excluded. RESULTS:Fourteen participants completed the study (forced expiratory volume in 1 s = 35 ± 10% predicted). Compared with ACS, SpO(2) measured via the finger sensor was 2% lower (limit of agreement 3%), and SpO(2) measured via the forehead sensor was 2% higher (limit of agreement 4%). Differences were not systematic. The change in oxygen saturation during exercise was similar among the finger sensor (-7; 95% confidence interval (CI): -4 to -10%), forehead sensor (-7; 95% CI: -3 to -10%) and ACS (-6; 95% CI: -3 to -9%). CONCLUSIONS:Oxygen saturation measured using the forehead sensor was higher than that measured in ACS. Assuming that oxygen saturation in ACS is slightly less than arterial blood, forehead sensors may yield measures more concordant with arterial blood. Both sensors detected exercise-induced desaturation.