High-density lipoprotein (HDL) is highly heterogeneous in its size and composition. Till now, the link of HDL subfractions to coronary risk is less clear. We aimed to investigate the associations of HDL subfractions with traditional risk factors (RFs), coronary severity, and outcomes in a cohort of nontreated patients with stable coronary artery disease (CAD). We prospectively enrolled 591 eligible patients. Baseline HDL subfractions were separated by Lipoprint system. HDL subfractions (large, medium, and small) and HDL-cholesterol (HDL-C) levels were dichotomized into low and high group according to the 50 percentile. Coronary severity was evaluated by SYNTAX, Gensini, and Jeopardy scoring systems. Patients were followed up annually for major adverse cardiovascular events (MACEs). Cox proportional hazards' models were used to evaluate the risk of HDL subfractions on MACEs. Patients with high large HDL-C levels had a decreased number of RFs. Significantly, large HDL-C levels were negatively associated with coronary severity assessed by SYNTAX and Gensini score (both P < 0.05). New MACEs occurred in 67 (11.6%) patients during a median 17.0 months follow-up. Moreover, the log-rank test revealed that there was a significant difference between high and low large HDL-C groups in event-free survival analysis (P = 0.013), but no differences were observed in total HDL-C groups and medium or small HDL-C groups (both P > 0.05). In particular, the multivariate Cox-proportional hazards model revealed that high large HDL-C was associated with lower MACEs risk (hazard ratio [95% confidence interval] 0.531 [0.295-0.959]) independent of potential confounders. Higher large HDL-C but not medium, small, or total HDL-C is associated with lower cardiovascular risk, highlighting the potential beneficial of HDL subfractionation.