The results of an experimental evaluation of Bartlett, Capon algorithms and cross-correlation function method effectiveness for direction finding of unmanned aerial vehicles are presented. Full-scale studies were carried out using a harmonic signal and acoustic radiation of a prototype of a propeller-driven UAV group indoors and in open space, as well as in real-world conditions of flight modes and UAV hovering. The acoustic emission spectrum of the electric propeller-motor group of modern UAVs includes narrow-band tonal and broadband noise-like components with predominant propeller radiation. The use of algorithms for determining the direction of arrival of acoustic radiation from the spatial spectrum by the methods of Bartlett and Capon involves the use of narrow-band signals. It is shown that to work with UAV acoustic radiation, the MB, MK, and cross-correlation function algorithms require some adaptation (including the use of high-pass filters). The Capon method has a significantly higher resolution than the classic Bartlett method, and a lower level of side lobes. The presence of significant reflections from local objects leads to the appearance of abnormal bearing estimates for the acoustic signal source of the propeller-driven UAV group. The results of determining the bearing of the source of acoustic radiation in a wide frequency band in open space show good agreement with the specified values of the bearing when applying the method of mutual correlation function. To increase the efficiency of the mutual correlation function algorithm when processing UAV acoustic signals, it is necessary to reduce the influence of low-frequency interference caused by anthropogenic and natural acoustic noise.