In the present work, a methodology for multiproduct quantification of paracetamol, caffeine and sodium diclofenac using ultraviolet (UV) detection (first order data) associated with multivariate curve resolution with alternating least squares (MCR-ALS) and Partial Least Squares (PLS), was developed. Generally, for first order data modelling, a high number of samples are required to include all the constituents present in future samples. However, it is possible that interferences impair the use of the calibration model when it comes to multiproduct analyses. We propose a methodology based on the use of individually built calibration curves, one for each analyte, similarly to a univariate calibration, and use chemometrics approaches to accurately quantify the analytes in the presence of interferences. Two strategies were evaluated: 1) the individual curves were grouped in a calibration matrix and 2) only the curve of the analyte of interest was used for calibration. These strategies were applied to two sample sets: 1) a test set prepared with a mixture of the analytes, according to a Central Composite Design and 2) 15 commercial drug products with very different compositions and dosage forms. Partial Least Squares (PLS) and Multivariate Curve Resolution with Alternating Least Squares (MCR-ALS) were compared and are discussed regarding first and second order advantages in the two approaches. PLS presented adequate performance using strategy 1 and the test set mixture, indicating that there was no need to prepare mixtures of the analytes for calibration. MCR-ALS achieved accurate results for both strategies and both sample sets, fulfilling the requirements to achieve first and second order advantages, which are related to the capability of providing the analyst with information about the presence of unmodelled compounds in unknown samples, on the one hand, and achieving an accurate quantification in the presence of these unknown compounds, on the other hand. Building a calibration set with standards of pure compounds not only will reduce the number of samples required for calibration, but makes it easier to update the model by simply including new calibration samples for the quantification of new compounds. This is an interesting strategy for pharmaceutical analysis, since the composition of samples is known by the manufacturer Ver menos