Tree-ring-based climate reconstructions are typically derived from either (1) a single species from one or multiple locations or (2) multiple species from multiple locations. Here, we investigate the ability of using multiple co-occurring canopy-dominant species from a single location for climate reconstructions based in the eastern United States. Using a variety of techniques, we first compare the climate signals of three canopy-dominant species (Quercus rubra, Quercus alba, Liriodendron tulipifera) at an old-growth forest in southern Indiana. We then determine if a composite time series of these co-occurring species increases or decreases the reconstruction model skill. Climate–growth correlation analysis of the species reveals strong and consistent relationships with summer [June–August (JJA)] Palmer Drought Severity Index (PDSI) during the period 1895–2012. We first use a split-sample reconstruction technique to compare the performance of species reconstruction models. We then use a nested technique to build a composite chronology against which to compare the individual species chronologies. The composite chronology of all three species accounts for up to 38% of the mean variation in JJA PDSI, and verification statistics indicate robust statistical skill from 1870 to 2012. The composite chronology reconstruction also outperforms each individual species model, indicating that using multiple co-occurring species increases reconstruction skill, at least from a single study site. Furthermore, model performance is improved by using nested reconstruction techniques, and implicates the potential ability to use multiple co-occurring species across multiple locations in the eastern United States.