Capability for identification of direction of movement of sound images (upward or downward) was studied in listeners of two age groups: 19- 27-year-old (11 subjects) and 55- 73-year-old (9 subjects). Various sound models of movement in the median plane were used as stimuli. Initially, a model of movement was developed based on filtration of broadband noise pulses by sets of non-individualized. i. e., measured in other listeners, head-related transfer functions. These functions corresponded to consecutive positions of the sound sources with the 5.6 degrees step between the space points with coordinates of elevation from -45 degrees to 45 degrees. The signals generated on the basis of transfer function sets of 23 subjects were distinguished by regulatory and value of the spectral minimum shift as well as by dynamic changes of the spectral maximum. Then the dynamic changes typical on non-individual sound images were used to create synthetic models. In these models, movement of sound images in the median plane was simulated either by a regular shift of the spectral minimum in broadband noise pulses or by a combination of the spectral minimum shift with a simultaneous change of the spectral maximum width and power. The obtained data have shown that young listeners with the high capability for vertical localization could identify direction of the sound image movement based on displacement of the spectral minimum in the broadband noise. For identification of direction of the sound image movement, the younger listeners with poor capabilities for vertical localization and the older listeners used dynamic changes of the signal power, which were connected mainly with the spectral maximum range.