Quantitative structure-activity relationships (QSARs) are developed for two separate families of sweet-tasting molecules for which sweetness values relative to sucrose (RS) have been measured. For these two families of sucrose and guanidine derivatives, the molecules were divided into training and test sets. Linear multiple regression equations have been generated to relate separately log(RS) to two types of parameters, namely molecular descriptors and energies derived via molecular field analysis (MFA). The parameters used in the development of linear multiple regression equations were selected by the genetic algorithm. The equations obtained show high predictive quality, which is confirmed by statistical parameters obtained with the test sets.The data for these two families were then combined with data from two other families previously studied, namely the sulfamates and isovanillates, to make a set of 149 compounds. These molecules were also studied by QSAR methods. The generated equations show remarkable predictive power, and the quality of the results suggest that the mechanism of sweet taste receptor is similar and, therefore, that there could well be only one receptor site for sweet taste, particularly for the four sweet taste families considered in this work.