Multisensory integration of spatial signals requires not only that stimulus locations be encoded in the same spatial reference frame, but also that stimulus locations be encoded in the same representational format. Previous studies have addressed the issue of spatial reference frame, but representational format, particularly for sound location, has been relatively overlooked. We discuss here our recent findings that sound location in the primate inferior colliculus is encoded using a "rate" code, a format that differs from the place code used for representing visual stimulus locations. Possible mechanisms for transforming signals from rate-to-place or place-to-rate coding formats are considered.