Areas of resonance signals present in 12 frequency bins of the H-1 NMR spectra of the vacuum residues of five Colombian crudes and a blend were correlated with their SARA components and some physicochemical properties. The residues studied were obtained from light paraffinic to heavy aromatic Colombian crudes. Correlations between the NMR spectra and properties were obtained with partial least square PLS and multiple linear regression MLR. Prediction models for SARA obtained by PLS had R-2 > 0.97 while CV - q(2) was between 0.92 and 0.99. Models obtained with MLR for SARA showed a high R-ad(2) for the first three components and a lower one for asphaltene. The R-2 of the prediction models for density, API Gravity, S, N, Ni ppm and wax %m content plus nC5- and -C7 insolubles were >0.988 while CV - q(2) was between 0.87 and 0.99. The proposed correlation methods are fast and less costly than the conventional ones. (C) 2009 Elsevier Ltd. All rights reserved.