The objective of this study was to evaluate a new reflectance pulse oximeter sensor. The prototype sensor consists of 8 light-emitting diode (LED) chips (4 at 665 nm and 4 at 820 nm) and a photodiode chip mounted on a single substrate. The 4 LED chips for each wavelength are spaced at 90-degree intervals around the substrate and at an equal radial distance from the photodiode chip. An optical barrier between the photodiode and LED chips prevents a direct coupling effect between them. Near-infrared LEDs (940 nm) in the sensor warm the tissue. The microthermocouple mounted on the sensor surface measures the temperature of the skin-sensor interface and maintains it at a preset level by servoregulating the current in the 940-nm LEDs. An animal study and a clinical study were performed. In the animal study, 5 mongrel dogs (weight, 10–20 kg) were anesthetized, mechanically ventilated, and cannulated. In each animal, arterial oxygen saturation (SaO₂) was measured continuously by a standard transmission oximeter probe placed on the dog's earlobe and a reflectance oximeter sensor placed on the dog's tongue. In the first phase of the experiment, signals from the reflectance sensor were recorded while the dog was immersed in ice water until its body temperature decreased to 30°C. In the second phase, the animal's body temperature was normal, and the oxygen content of the ventilator was varied to alter the SaO₂. In the clinical study, 18 critically ill patients were monitored perioperatively with the prototype reflectance sensor. The first phase of the study investigated the relationship between local skin temperature and the accuracy of oximeter readings with the reflectance sensor. Each measurement was taken at a high saturation level as a function of local skin temperature. The second phase of the study compared measurements of oxygen saturation by a reflectance oximeter (SpO₂[r]) with those made by a co-oximeter (SaO₂[IL]) and a standard transmission oximeter (SpO₂[t]). Linear regression analysis was used to determine the degree of correlation between (1) the pulse amplitude and skin temperature; (2) SpO₂(r) and SaO₂(IL); and (3) SpO₂(t) and SaO₂(IL). Student'st test was used to determine the significance of each correlation. The mean and standard deviation of the differences were also computed. In the animal study, pulse amplitude levels increased concomitantly with skin temperature (at 665 nm,r=0.9424; at 820 nm,r=0.9834;p<0.001) and SpO₂(r) correlated well with SaO₂(IL) (r=0.982; SEE=2.54%;p<0.001). The results of the clinical study are consistent with these findings. The proto-type reflectance pulse oximeter sensor can yield accurate measurements of oxygen saturation when applied to the forehead or cheek. It is, therefore, an effective alternative to transmission oximeters for perioperative monitoring of critically ill patients.