Spectral information and wet analysis results were statistically evaluated by means of standard error values. For that purpose, the calibration and subsequent validation of the predictive models were performed, applying scattering correction and using different region of the spectra. Considering the standard error of calibration (SEC) and cross validation (SECV), a suitable method for the determination of different cannabinoids was accomplished. The residual predictive deviation (RPD) statistics were considered to evaluate the cannabinoids prediction, resulting in good prediction models for Δ9-THC, CBC and CBD. In the case of CBN, Δ8-THC, and CBDV, their models may be used for screening purposes while CBG and Δ9-THCV equations are not yet suitable to be used. Therefore, additional samples should be analyzed to increase the population enhancing the prediction models. The comparison of the predictive ability of the models obtained with the dispersive NIR and the FT-NIR spectrometers show no significant differences between them.