Background Dyslipidemia contributes to the development and progression of arterial stiffness. We aimed to identify the most informative measures of serum lipids and their calculated ratios in terms of arterial stiffness progression risk. Methods Total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and brachial-ankle pulse wave velocity (baPWV) of 659 healthy males (47.4 +/- 10.7 years) were measured at baseline. Values for non-HDL-C, TC/HDL-C, TG/HDL-C, LDL-C/HDL-C, and non-HDL-C/HDL-C were calculated. BaPWV was re-performed after 4.1 years follow-up. Elevated baPWV was defined as baPWV >= 1400 cm/s. Results Over the follow-up period, the mean baPWV value increased from 1340 cm/s to 1410 cm/s, and 331 individuals increased/persisted with high baPWV (outcome 1). Among the 448 subjects who had normal baseline baPWV, 100 incident elevated baPWV occurred (outcome 2). Only baseline logTG (OR 1.64 [95% CI: 1.14-2.37] for outcome 1; 1.89 [1.14-3.17] for outcome 2) and logTG/HDL-C (1.54 [1.15-2.10] for outcome 1; 1.60 [1.05-2.45] for outcome 2) were significantly associated with arterial stiffness progression after adjusting for confounding factors. Adding logTG or logTG/HDL-C to age and blood pressure improved the accuracy of risk predictions for arterial stiffness progression. These associations remained significant when lipids were analyzed as categorical variables. Conclusions Baseline serum TG and TG/HDL-C were independently associated with increases in/persistently high baPWV and incident elevated baPWV, and they performed more effectively than other lipid variables in identifying healthy men at high risk of arterial stiffness progression.