Some lipoprotein disorders related to the residual risk of premature cardiovascular disease (PCVD) are not detected by the conventional lipid profile. In this case-control study, the predictive power of PCVD of serum sdLDL-C, measured using a lipoprotein precipitation method, and of the physicochemical properties of serum lipoproteins, analyzed by nuclear magnetic resonance (NMR) techniques, were evaluated. We studied a group of patients with a first PCVD event (n = 125) and a group of control subjects (n = 190). Conventional lipid profile, the size and number of Very Low Density Lipoproteins (VLDL), Low Density Lipoproteins (LDL), High Density Lipoproteins (HDL) particles, and the number of particles of their subclasses (large, medium, and small) were measured. Compared to controls, PCVD patients had lower concentrations of all LDL particles, and smaller and larger diameter of LDL and HDL particles, respectively. PCVD patients also showed higher concentrations of small dense LDL-cholesterol (sdLDL), and triglycerides (Tg) in LDL and HDL particles (HDL-Tg), and higher concentrations of large VLDL particles. Multivariate logistic regression showed that sdLDL-C, HDL-Tg, and large concentrations of LDL particles were the most powerful predictors of PCVD. A strong relationship was observed between increased HDL-Tg concentrations and PCVD. This study demonstrates that beyond the conventional lipid profile, PCVD patients have other atherogenic lipoprotein alterations that are detected by magnetic resonance imaging (MRI) analysis.