I-Mutant2.0 is a support vector machine (SVM)-based tool designed to predict protein stability changes upon single point mutations. It can predict both the direction (ΔΔG sign) and the magnitude (ΔΔG values) of these changes, either from the protein structure or sequence. The method was trained and tested on a dataset from ProTherm, the most comprehensive database of thermodynamic experimental data for protein stability changes under different conditions. I-Mutant2.0 achieves 80% accuracy when predicting the direction of stability changes using structural information and 77% when using sequence information. For predicting ΔΔG values, the correlation between predicted and experimental values is 0.71 (standard error 1.30 kcal/mol) and 0.62 (standard error 1.45 kcal/mol) for structural and sequence-based predictions, respectively. The web interface allows users to select between structure- and sequence-based predictions, making it a valuable tool for protein design, even when the protein structure is not available.I-Mutant2.0 is a support vector machine (SVM)-based tool designed to predict protein stability changes upon single point mutations. It can predict both the direction (ΔΔG sign) and the magnitude (ΔΔG values) of these changes, either from the protein structure or sequence. The method was trained and tested on a dataset from ProTherm, the most comprehensive database of thermodynamic experimental data for protein stability changes under different conditions. I-Mutant2.0 achieves 80% accuracy when predicting the direction of stability changes using structural information and 77% when using sequence information. For predicting ΔΔG values, the correlation between predicted and experimental values is 0.71 (standard error 1.30 kcal/mol) and 0.62 (standard error 1.45 kcal/mol) for structural and sequence-based predictions, respectively. The web interface allows users to select between structure- and sequence-based predictions, making it a valuable tool for protein design, even when the protein structure is not available.