The objective of this study was to evaluate the potential of near-infrared (near IR) spectroscopy associated with multivariate statistics to distinguish charcoal produced from wood of planted and native forests in Brazil. Timber forest species from the Cerrado (Cedrela sp., Aspidosperma sp., Jacaranda sp., and Apuleia sp.) and Eucalyptus clones from forestry companies (Vallourec steel producer and Cenibra pulp producer) were pyrolyzed under well controlled laboratory scale conditions at the final temperatures of 300 (573,15), 500 (773,15) and 700 °C (973,15 K), respectively. Fifteen charcoals of each species were produced for each temperature leading to heighten controlled pyrolysis treatments and finally 270 charcoal samples (3 treatments × 15 repetitions × 6 materials). Principal component analysis (PCA) and partial least-squares regression (PLS-R) were carried out in the spectra recorded from charcoal specimens. Near IR spectroscopy associated with PCA was not able to differentiate the charcoals produced from native and planted woods if the 270 samples were considered in the same analysis. However, the separation of native and planted charcoal was achieved when the samples were analyzed separately by final pyrolysis temperature. Thus, the prediction of native or planted classes by PLS-R presented better performance for samples pyrolyzed at 300 °C, followed by those at 500 °C, 700 °C, and all together.