Geographic variation in the structure of corncrake calls was examined by the analysis of recordings collected in four areas of the species' range. On a large geographic scale (> 100 km), discriminant function analysis correctly attributed 54-88% of individuals to the population from which they were recorded based upon temporal measures of call structure. On a smaller geographic scale (< 1 km), neighbouring individuals shared significantly more similar call structures than more distant individuals. Mantel tests showed a high level of clustering of similar call structures at three sites within the areas studied. The implications of these results for conservation efforts are discussed.