Mammalian sensory cortex is functionally partitioned into cortical fields that are specialized for different processing operations. In theory, averaging functional and anatomical images across subjects can reveal both the average anatomy and the mean functional organization of sensory regions. However, this averaging process must overcome at least two obstacles: (1) the relative locations and sizes of cortical sensory areas vary in different subjects so that across-subject averaging introduces spatial smearing; (2) the relative locations and sizes of cortical areas vary between hemispheres, making it difficult to compare activations between hemispheres or to combine activations across hemispheres. These difficulties are particularly acute for small cortical regions such as auditory cortex. In whole-brain averaging procedures, considerable intersubject variance in the location and orientation of auditory cortex is introduced by variance of the size and shape of structures outside auditory cortex. Here, we compared these global methods with local landmark-based methods (LLMs) that use warping based on local anatomical landmarks. In comparison to maps made with global methods, LLMs produced anatomical maps of auditory cortex with clearer gyral and sulcal structure, and produce functional maps with improved resolution. These results suggest that LLMs have significant advantages over global mapping procedures in studying the details of auditory cortex organization.