In the quality control of medicines and the fight against the phenomenon of poor quality medicines, there is an urgent need for rapid and broad spectrum methods for screening these types of medicines. In the present work, we have used near infrared spectroscopy combined with multivariate data analysis to develop chemometric models for the classification and quantification of metronidazole in Burkina Faso pharmaceutical formulations. For this purpose, drug samples were collected in drugstores located in different Burkina Faso border zones. Four product classes were defined based on the national nomenclature: 3 classes for the generic drugs (C1, C3, and C4) and one class for the reference (C2) drugs. The exploratory analysis using PCA identified two clusters of drugs within class C1. Discrimination was confirmed by the developed and optimised DD-SIMCA model, with only one target class. The quality control of the samples from product class C1 was proven to be very satisfactory with specificities and sensitivities of 100%. The quantification models developed with the PLS-R method were successfully applied for the determination of the active ingredient content in the samples, with acceptable relative bias between 0.15 and 12.7 % with respect to the dose determined by the HPLC method. The RMSEC was estimated at 13.57 (R2, 0.9937), the RMSECV at 18.07 (R2, 0.9888) and the RMSEP at 13.69 (R2, 0.9941).The models developed and the results obtained are promising for routine quality control of similar formulations of metronidazole.