Quantitative analysis based on near-infrared (NIR) spectroscopy uses individual calibration equations for each sample studied because of the need to model all possible chemical and physical variability sources. A NIR method, using a fibre optical probe, for the analysis of two different and related pharmaceuticals from two different production steps (cores and tablets) is studied. Both pharmaceuticals have the same active principle and similar excipients, but with different nominal concentration values. The pharmaceuticals are identified by comparison using a second-derivative spectral library; an identification procedure consisting of two-steps (cascade) library: correlation coefficient followed by maximum distance in wavelength space is proposed. Once a sample has been positively identified, the active principle is quantified with partial least-squares regression (PLSR) using a sole and global calibration. The proposed method was validated for use as a control method, and for this purpose selectivity of the identification process, and the repeatability, intermediate precision, accuracy, linearity, and robustness of the active principle quantitation, were assessed. (C) 2000 Elsevier Science B.V. All rights reserved.