Auditory localisation was examined using brief broadband sounds presented during rapid head turns to visual targets in the peripheral field. Presenting sounds during a rapid head movement will "smear" the acoustic cues to the sound's location. During the early stages of a head turn, sound localisation accuracy was comparable to a no-turn control condition. However, significant localisation errors occurred when the probe sound was presented during the later part of a head turn. After correcting for head position, the estimate of lateral angle (horizontal position) in the front hemisphere was generally accurate. However, lateral angle estimates for positions in the rear hemisphere exhibited systematic errors that were especially large around the midline. Polar angle (elevation) perception remained robust, being comparable to no-turn controls whether tested early or late in the head turn. The results are interpreted in terms of a 'multiple look' strategy for calculating sound location, and the allocation of attention to the hemisphere containing the head-turn target.