A quantitative structure-activity relationship analysis was employed to explore the relationship between the molecular structure of thiosemicarbazone analogues and the inhibition of the cysteine protease cruzain, a validated target for Chagas' disease treatment. A data set containing 53 thiosemicarbazone derivatives was used to produce a quantitative model for activity prediction of unknown compounds. Several electronic descriptors were obtained through DFT calculations, along with a large amount of Dragon descriptors. The ordered predictor selection (OPS) algorithm was employed to select the most relevant descriptors to perform PLS regressions. With this procedure, significant correlation coefficients (r(2) = 0.85, q(2) = 0.78) were achieved. Furthermore, predicted values for an external test set are in good agreement with the experimental results, indicating the potential of the model for untested compounds. Additional validation tests were carried out, indicating that a robust and reliable model was obtained to be used in the design of new thiosemicarbazones with improved cruzain inhibition potential.