The sweetness values relative to sucrose (RS) of a series of sweet-tasting guanidine derivatives have been modeled using augmented multivariate image analysis applied to quantitative structure–property relationship (aug-MIA-QSPR). This QSPR approach is based on 2D molecular shape, as well as atomic sizes and colors to encode chemical, physical and biological properties. A 2 predictive model with r 2 = 0.955, q 2 = 0.731 and r test 1⁄4 0:571 was used to estimate log(RS) of two novel sweet- tasting guanidine derivatives, which were built from the combination of substructures of the sweetest derivatives along the series. Given the synergistic effect of different substituents to provide new molecules, promising guani- dine sweeteners are proposed for further synthesis.