A simple quantitative structure–activity relationship (QSAR) method of analysis used to predict biological activity for congeneric series of compounds is reported. This method is based on the application of bilinear or multilinear partial least squares regression to a data set, which is a binary matrix representing the substituents of a framework. It is appraised here to a series of (S)-N-[(1-ethyl-2-pyrrolidinyl)methyl]-6-methoxybenzamides, compounds with affinity towards the dopamine D 2 receptor subtype and showed high predictive ability, even when compared to a refined three-dimensional (3D) approach.