H-1 NMR spectra and transversal relaxation time (T-2) distributions of 76 Brazilian crude oil samples, ranging from light to extra-heavy, were measured and compared by Hierarchical Clustering Analysis (HCA) and Principal Component Analysis (PCA). Partial Least Squares Regression (PLSR) models were developed to predict crude oil viscosity and density in terms of the American Petroleum Institute (API) gravities derived from their individual NMR responses. Relaxation-based models delivered better predictive performances for both viscosity and API, with a standard error of prediction (SEP) of 0.220 log cP and 1.027 degrees API. In addition, the synergistic effect between the two distinct NMR responses was evaluated for the first time by merging spectra and relaxation data into a single PLSR model. This hybrid spectrarelaxation model delivered superior performance compared to stand-alone models, with an SEP of 0.186 log cP and 0.811 degrees API. (C) 2014 Elsevier Ltd. All rights reserved.