Oxygen in arsenic and various arsenic compounds was determined by Schutzo-Unterzaucher method. When the determination of oxygen in arsenic compounds was repeated 5 or 6 times by the method using platinum-carbon, the activity of platinum-carbon deteriorated. The analytical values of arsenic trioxide were less than the theoretical value by about 1%. When nickel-carbon was used in place of platinum-carbon, the deterioration was less than that of platinum-carbon. When the determination of oxygen in arsenic trioxide was repeated 17 times, the analytical values of sucrose and phenolphtalein were less than the theoretical value by about 1%. However, it was not advisable to use nickel-carbon for the routine analysis. The experiment using two-layer packing of either carbon-platinum-carbon or nickel-carbon-nickel-carbon did not give good results. Determination using carbon pretreated with hydrogen and heated at above 1030° was found possible with an error of 0.2%. One analysis required approximately 35 min.