Time domain nuclear magnetic resonance (TD-NMR) and chemometric models were used to classify intact plums according to the total soluble solid content in percentage (% of SSC). The models were constructed using use Carr–Purcell–Meiboom–Gill (CPMG) decay (as independent variables) of intact plums. The reference values in % of SSC (as dependent variable) were obtained from the juice extracted from the same plums. The data sample was divided into two categories: class 1, up 9 to 12% of SSC and class 2, up 13 to 22% of SSC. The soft independent modeling of class analogy (SIMCA) of 171 plums correctly classified the class 1 plums approximately 89% of the total and the class 2 plums (sweeter) above 93% of the samples for both training and the validation data set, at 95% of confidence level. Therefore, the analysis of CPMG decay using SIMCA can be used to classify intact plums according to sweetness. This method is fast (seconds) and can easily be automated, allowing a better classification of each plum than the average SSC values of a lot.Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)