A series of experiments compared short-term memory for object locations in the auditory and visual modalities. The stimulus materials consisted of sounds and pictures presented at different locations in space. Items were presented in pure- or mixed-modality lists of increasing length. At test, participants responded to renewed presentation of the objects by indicating their original position. If two independent modality-specific and resource-limited short-term memories support the remembering of locations, memory performance should be higher in the mixed-modality than in the pure-modality condition. Yet, memory performance was the same for items in both types of list. In addition, responses to the memory load manipulation in both modalities showed very similar declines in performance. The results are interpreted in terms of object files binding object and location information in episodic working memory, independently of the input modality.