Different indexes have been proposed in the literature to measure the stability and compatibility of crude oil blends, including the insolubility number (I-N), which measures the degree of asphaltene insolubility, and the solubility blending number (S-BN), which measures the ability of the oil to dissolve asphaltenes. In this work, various chemometric models were developed to predict the I-N and S-BN values of Colombian heavy crude oils (degrees API from 6 to 27), in which the integral areas of the resonance signals, from 12 regions of H-1 nuclear magnetic resonance (H-1 NMR) spectra, were correlated with their I-N and S-BN. Correlations between the H-1 NMR spectra and the said properties were found via partial least squares (PLS) regression so as to create the predictive models. Coefficients of determination (R-2) above 0.92 and cross-validated (CV) predictive correlation coefficients (q(cv)(2)) above 0.86 were attained with the developed PLS prediction models for I-N and S-BN. The use of these NMR-based predictive methods entails a faster estimation of the stability and compatibility of crude oil blends and a more eco-friendly and cheaper methodology compared to conventional methods. From the H-1 NMR data, it is observed that crude oils with a low tendency to precipitate asphaltenes (high S-BN) are those with a high aromatic content and a low content of paraffins.