Experiments on the localization of air-borne sound were conducted in an anechoic room employing an acoustical "pointer" as the subject''s method of indicating the direction of the stimulus tone. The pointer was a small loudspeaker carried on a boom which rotated about a vertical axis through the subject''s head. This speaker presented a wide-band noise which alternated with the tone to be localized. The switching was transientless and was performed by an electronic gate having a 100-millisecond rise and decay time. Three small loudspeakers in enclosures, one mounted directly in front of the subject and one 40[degree] to each side, presented the stimulus tones to be localized. In the 1st experiment the speakers were employed singly. In the other two experiments they were used in pairs. In the 2d experiment the pairs of speakers were in phase; in the 3d, they were in phase opposition. The stimulus conditions of Experiment 2 generated a "phantom source" which appears to lie between the two speakers employed. The predicted location was compared with the subjects'' responses. The stimulus conditions of Experiment 3 generated a phantom source which according to prediction, should lie toward the side opposite to the asymmetrically placed speaker and should move in direction with frequency. This prediction was borne out by the subjects'' responses for frequencies where interaural time was the dominant basis for localization. Results of the 3 experiments support the conclusion that the localization of tones below about 1500 cps is determined largely by interaural time differences. For frequencies above 1500 cps, differences of intensity at the two ears must be the dominant factor, but neither probe microphone measurements nor audiograms for the subjects provided information from which satisfactory predictions could be made.