The accuracy of pulse oximeters in measuring transient changes in oxygen saturation (SaO2) may be affected by the oximeter time response. The aim of this study was to assess the effect of modifying the pulse oximeter averaging time (T) on the measurement of SaO2 in patients with the sleep apnea-hypopnea syndrome (SAHS). Twelve patients with severe SAHS were studied during a nap with conventional oximeters: Ohmeda 3740 and Criticare 501. We compared the readings of each patient's oxygen desaturation measured simultaneously with two identical pulse oximeters. One oximeter was the control (T = 3 seconds), and in the other T was set from 3 seconds to 21 seconds. No significant differences in SaO2 were found when both oximeters were set to the same T (3 seconds). In contrast, increasing T to 12 seconds and 21 seconds in one of the oximeters resulted in considerable and significant differences in the measured SaO2: oxygen desaturation was underestimated by up to 60% when compared with the control. The misestimation of SaO2 induced by settings of T which are within the range selectable in conventional oximeters may be of epidemiological significance when pulse oximetry is used as a complementary diagnostic tool to classify sleep events in SAHS.