AbstractSwitchgrass varieties grown under various environments were investigated by dispersive and Fourier Transform Near-Infrared (NIR) spectrometers. The collected NIR spectra were analyzed using multivariate approaches. More specifically, principal component analysis (PCA) and projection to latent structures (PLS) regression techniques were employed to classify and predict characteristics of the switchgrass samples. The multivariate results were compared to reflectance indices that are commonly used to study the physiological performance of plants. From near infrared spectra, discrimination between the two growth locations was successfully achieved by PCA. Separation based on the ecotype and the rate of fertilizer applied to the field was also possible by the multivariable analysis of the spectral data. For the classification/ discrimination of the switchgrass samples, the near infrared spectra collected by the dispersive and the Fourier Transform spectrometers provided similar results. From the two near infrared data sets robust models were developed to predict non-structural carbohydrates content and the rate of nitrogen applied to the field. However, the spectra collected by the dispersive spectrometer resulted in more accurate models for these samples.