Mahogany is one of the most valuable woods and was widely used until it was included in Appendix II of the Convention on International Trade in Endangered Species as endangered species. Mahogany wood sometimes is traded under different names. Also, some similar woods belonging to the Meliaceae family are traded as “mahogany” or as being of a “mahogany pattern”. To investigate the feasibility of the use of near infrared spectroscopy for wood discrimination, the mahogany (Swietenia macrophylla King.), andiroba or crabwood (Carapa guianensis Aubl.), cedar (Cedrela odorata L.), and curupixá (Micropholis melinoniana Pierre) woods were examined. Four discrimination models based on partial least squares-discriminant analysis were developed based on a calibration set composed of 88 samples and a test set with 44 samples. Each model corresponds to the discrimination of a wood species from the others. Optimization of the model was performed by means of the OPUS^® software followed by statistical analysis software (Matlab^®). The observed root mean square errors of predictions were 0.14, 0.09, 0.12, and 0.06 for discriminations of mahogany, cedar, andiroba, and curupixá, respectively. The separations of the species obtained based on the difference in the predicted values was at least 0.38. This makes it possible to perform safe discriminations with a very low probability of misclassifying a sample. This method can be considered accurate and fast.