Three migratory subspecies of sandhill crane (Grus canadensis) occur in North America: greater (G. c. tabida), Canadian (G. c. rowani), and lesser (G. c. canadensis). These subspecies vary clinally in size from the large tabida to the small canadensis. All 3 subspecies co-occur during the nonbreeding season, but field identification is challenging and census efforts typically do not even attempt to distinguish them. We developed a novel method to determine the subspecies composition of nonbreeding sandhill crane populations using formant spacing in vocalizations. Each note in a crane vocalization is comprised of several formants, or energy peaks in the frequency spectrum. Formant spacing is inversely proportional to the length of the sound-emitting tube, or trachea. Analysis of formant-spacing distributions for tabida and canadensis revealed that tabida has reduced formant spacing, as predicted based on its larger body size and correspondingly larger trachea. Comparisons of these subspecies-specific formant-spacing distributions to formant-spacing distributions from calls recorded at 2 crane wintering areas in New Mexico, USA, showed that the wintering crane populations of the Middle Rio Grande Valley and the Lower Rio Grande Valley contain strikingly different proportions of the globally rare tabida (75.5% and 5.6%, respectively). These findings are concordant with crane subspecies composition estimates derived from hunter check stations in New Mexico. Formant-spacing provides an indirect, noninvasive method of estimating body-size distributions that has several practical advantages. We expect it to be most useful when body-size classes have limited overlap, as is the case with New Mexico sandhill crane populations. © 2012 The Wildlife Society.