Various methods were developed to predict the stability of Colombian crude oils, in which the integral areas of the resonance signals from 12 regions of H-1 nuclear magnetic resonance (NMR) spectra of 29 widely different crude oils were correlated with the stability parameter of Heithaus (P-o) and the colloidal instability index (CII). Correlations between the NMR spectra and properties were obtained using partial least-squares (PLS) regression and multiple linear regression (MLR). The prediction models for P-o and CII by PLS had coefficients of determination (R-2) of >98 and >99%, respectively, while the cross-validation values (CV, q(2)) ranged from 89 to 90%, respectively. The models obtained from MLR showed a high adjusted R-2 (R-ad(2)) for P. and a lower R-ad(2) for CII. The R-2 values of the prediction models for P-o ranged from 97 to 98%. The use of these predictive methods is faster, more environmentally friendly, and less expensive than conventional methods. Of the set of crude oils used in this study, it was observed that the crude oils with a low tendency to precipitate asphaltenes are those with a high aromatic content and low paraffin content because they exhibited a very low CII and a very high peptizing power for asphaltenes, P-o. Considering the relationship between the asphaltene content and P-o and CII, asphaltenes cannot be considered negative factors for the stability of some crude oils.