We asked whether geographic variation exists in the complexity of song repertoires in Song Sparrows (Melospiza melodia) by quantitatively comparing four measures of repertoire organization across four geographically distant populations: (1) repertoire size (the number of distinct song types), (2) the number of 'minimal units of production' per repertoire, (3) mean similarity among variants of the same song type ('within-type' similarity), and (4) mean similarity among song types in a repertoire ('between-type' similarity). We found significant geographic differences among populations in three of these four measures, with mean similarity among song types being the exception. In general, relatively sedentary populations in North Carolina and Washington were more similar to each other than to migratory populations in Pennsylvania and Maine. Contrary to our expectation based on prior interspecific analyses of variation in repertoire complexity, the relatively sedentary populations in our sample had more complex repertoires than did the more migratory populations. The origin and functional significance of population differences in repertoire complexity in this species remain uncertain.