Three experiments are presented comparing the effectiveness of several parameters of sound for the auditory presentation of statistical data. The dimensions of pitch, loudness, panning, and time were used alone and redundantly to map the values of a box plot to an auditory display. Temporal mappings resulted in better performance than mappings using pitch, panning, or loudness. In the first two experiments, there was no benefit when the mapping condition used two dimensions redundantly over mappings using one dimension. However, for the third experiment, there was a benefit of a redundant design when the dimensions of sound used were integral whereas there was no benefit when they were separable. This third experiment used a task more closely approximating a real-life application of auditory displays. Its results suggest that sonification can be used effectively in situations requiring the monitoring of more than one source of information.