The aim of this study was to propose an alternative methodology to determine the total antioxidant in curcumin. To this, fresh curcumin was dried, ground and the resulting powder was submitted to extractions based on a 23 full factorial design. The factors, evaluated in two levels, were ethanol concentration (70 and 99.8%); drying type (stove with air circulation and freeze drying) and time (2.5 and 5 minutes). The obtained results were expressed in terms of total concentration of curcuminoids. The results suggest that the drying type based on stove with air circulation, ethanol concentration 70% and time in upper level promotes a more effective extraction. From this extract, samples were prepared from 2mL of distillated water with 25, 50, 75, 100 and 150mL of extract. Near Infrared (NIR) spectra were collected and the antioxidant activity were determined based on 2,2-diphenyl-1-picrylhydrazyl (DPPH) method. Multivariate Curve Resolution with Alternating Least Squares (MCR-ALS) chemometric tool was applied to the NIR spectra in order to promote the pure spectra separation related to the antioxidants and its relative concentrations. The relative concentrations related to the antioxidants were employed to build a pseudo-univariate calibration model against the antioxidant results obtained from DPPH method. The correlation coefficient for the model was 0.9532. By employing this model to predict the concentration of a sample, it was observed that by estimating the concentration of a reference sample (with known concentration of 90.26%), the result obtained was 87.50%. The results suggest that the combination of NIR measurements and MCR-ALS makes possible the direct determination of antioxidant activity in matrices with unknown interferents. Furthermore, the proposed methodology present the advantage of being free of sample preparation and the possibility to determine concentrations in the presence of unknown interferents.