This paper presents a novel method of real-time collision detection in virtual environments based on spherical harmonics and distance fields. Fundamentals of spherical harmonics modeling virtual object are initially proposed for representing complex objects with different degrees of spherical harmonics in the spherical coordinate system. Distances between virtual objects and space mesh are then projected onto the spherical surface to generate a distance map. Approximate collision detection is performed at the low scale, while accurate detection is performed at the high scale. Experimental results show that computational complexities and memory requirements are reduced significantly compared with those encountered in, for example, SQ-map method and the adaptive distance field, due to the fact that in this case the bounding sphere and the virtual object represented by spherical harmonics are in the same spherical reference system.