Music is one aspect of acoustical environment. According to the development of technology, the music production processes and the listening situations have been considerably diversified. Recently we have had many opportunities to listen to music by electro-acoustical reproduction systems. The reproduced musical sounds are supposed to reflect the listener's desire on music. In the present study of the acoustical environment of music, the optimum listening volume of the reproduced music was measured by the psycho-acoustical experiment using a method of adjustment. A-weighted equivalent sound level was used as a criterion of loudness. Further, the musical impression was evaluated by a semantic differential method using 15 scales. The average power spectrum and the temporal sound pressure level fluctuation were measured as the acoustical properties of musical sounds. The powerful and sharp music is listened at a very large volume. The loud music usually shows a trapezoidal spectrum pattern with strong high-frequency components and a smaller fluctuation pattern of sound pressure level fluctuation. On the contrary, the soft music usually shows a triangular spectrum pattern and a larger level -fluctuation pattern.