The phenolphthalin method of Haagen-Smit and Brunelle (1958) was chosen for a preliminary survey of total oxidant level in Pretoria air, because of its sensitivity. Difficulty, however, was encountered in obtaining reliable standard curves. Some improvement was obtained when conducting all operations except photometer measurements at the temperature of melting ice. It was also found that when the sequence of adding the reagents was changed, so as to simulate conditions during actual sampling, a standard curve approximating a straight line and differing considerably from that of McCabe (1953) was obtained. It follows that values of total oxidant obtained by any experimentor will depend to a certain extent upon the method of standard curve preparation he uses, and when comparisons are made between measurements by experimentors in different towns or countries this factor should be taken into consideration. The accuracy (95% confidence) obtained by the phenolphthalin method, using the mean of three successive samples, was shown to be in the region of 30% for very low amounts of oxidant.