Non-high-density lipoprotein cholesterol (non-HDL-C) may be equivalent or superior to low-density lipoprotein cholesterol (LDL-C) for prediction of cardiovascular disease (CVD) risk. However, studies comparing the predictive values of LDL-C and non-HDL-C for CVD and total mortality in a long-term follow-up yielded conflicting results. The Cardiovascular Occupational Risk Factor Determination in Israel Study (CORDIS) is a prospective cohort study of a young industrial population of workers with a long-term follow-up. The initial phase of the study was carried out in 1985-1999. Interviews and physical examinations were conducted, and fasting blood samples, including lipid panels, were undertaken. In 2007, after a 22-year follow-up period, the baseline data were merged with data on all-cause and CVD mortality obtained from the Israeli National Death Registry. A total of 4,832 men were included in the analysis with a mean age of 42.1 ± 12.1 years. Univariate analysis indicated a positive association between non-HDL-C and LDL-C levels and an increased risk for both all-cause and CVD mortality. Multiple regression analysis, following adjustment for potential confounders, resulted in attenuation of the association of both lipoproteins with total mortality. The adjusted association between non-HDL-C levels ≥190 mg/dl and CVD mortality remained significant (hazard ratio 1.80, 95% confidence interval 1.10 to 2.96), but the association of LDL-C with CVD mortality was attenuated (hazard ratio 1.53, 95% confidence interval 0.98 to 2.39). In conclusion, non-HDL-C may be a more potent predictor of CVD mortality than LDL-C levels.