Herbicide preparations, which typically consist of an active compound and an adjuvant, are formulated in such a way as to facilitate rapid dispersion of the pesticide and its transfer across the fat cuticle of plant leaves. The adjuvant usually comprises various substances including an oil, water, and one or more emulsifiers for increased stability. The difficulty of analyzing these mixtures has severely restricted the ability to control herbicide preparations, which must usually be characterized in terms of global, non-specific parameters such as the acid, saponification, and hydroxyl values. Near-infrared spectroscopy is an expeditious analytical technique requiring no sample preparation and affording multi-determinations in complex mixtures. In this work, we used laboratory-made samples to construct PLS multivariate models for determining the components of the adjuvant mixture in a herbicide preparation. The best ensuing model was validated and used to analyze industrial preparations with good results.