During the summers of I967 I968 and I969, I studied the acoustical communication system of the Indigo Bunting (Passerina cyanea) by means of playback experiments. Attention was directed to deciphering those properties of territorial song used in species identification. I tested the importance of different song parameters by exposing birds to recordings of normal and artificially modified vocalizations. The level of agonistic response exhibited by male territory holders was used as a bioassay of the effectiveness of the experimental song in allowing species recognition. The results indicate the following: I) Syntactical featues of song (the arrangement or order of notes (= figures) within the song) are not essential recognition cues. 2) Altering the rhythmic cadence of the song by lengthening or shortening the intervals between successive notes causes a marked reduction in male responsiveness. 3) The structure or morphology of the individual notes is also important in song identification. I hypothesize that notes must have a particular acoustical quality characterized by their covering a wide frequency range (mean = 4 kHz) in a short time interval (mean length = 0.18 seconds) and by their containing abrupt changes and reversals of pitch. A wide diversity of bunting note-types fit this description, and differences in details of their fine structure probably provide the variability necessary for permitting individual identification. 4) Only a small number of the acoustical cues potentially available are critical for species recognition. This high degree of feature extraction suggests that a) considerable redundancy is present in bunting song and, b) a large number of song parameters function in the conveyance of other (non-identification) messages. When these results are compared with data from other species of passerines (the Ovenbird, Seiurus aurocapillus, White-throated Sparrow, Zonotrichia albicollis, Wood Lark, Lullula arborea, and European Robin, Erithacus rubecula) several generalizations emerge: i) In all five species, recognition depends upon song features that are among the most constant and unvarying in the species repertoire. 2) Song characteristics that are constant in the repertoire of an individual bird but variable within a population are unimportant in species identification but often function in allowing recognition of specific individuals. 3) Those aspects of territorial song that are variable within an individuals's repertoire often convey information about the motivational state of the singer. While only a few vocal parameters are essential for identification in each of these species, the parameters serving this function cover the entire spectrum of morphological, temporal and syntactical features. One would like to be able to predict optimal physical structures for acoustical signals that convey recognition information. At present, too few studies have been conducted to make this possible. However, several speculations are advanced concerning the importance of behavioral and ecological factors such as population dispersion pattern, social organization, habitat structure, and the background sound environment.