The 'precedence effect' in sound localization refers to the situation in which two speakers, separated in space, emit identical sounds but one speaker leads by a few milliseconds. Normal observers perceive all the sound as originating from the leading speaker. In the present experiment cats were tested before and after unilateral ablation of auditory cortex on how well they could transfer a learned sound loclization response from situations involving unpaired sound sources to those with paired sources. Cats were tested with tone pairs separated by 5 msec and with clicks separated by delays of 3, 5, 7, 9, 12 and 16 msec. Preoperatively, the cats averaged 98% correct at a delay of 3 msec and 58% correct at 16 msec. Before surgery the probability of an error was independent of which speaker was leading; after surgery cats made significantly more errors when the leading speaker was located opposite the side of the lesion. Considerable individual variability was found with the 3 and 5 msec delay tests; some cats showed no evidence of a cortical deficit while others exhibited initial severe deficits which disappeared with training. At delays of 7-16 msec all cats showed severe directional deficits which were still present after 9 months of retraining. Results are interpreted as supporting the hypothesis that unilateral lesions disrupt binaural loudness rather than temporal relations.