Previous studies regarding effects of caffeine on lipids focused mainly on coffee consumption, the real association of serum caffeine concentrations with blood lipids is unclear. 814 caffeine-drug users who had taken any caffeine/caffeine-containing pharmaceutical products and 623 nonusers without any drug use in the last seven days before the medical interview were identified from German National Health Surveys from 1984 to 1999. Serum caffeine concentrations were measured by using EIA for caffeine-drug users and nonusers in the same laboratory. Blood lipids were measured routinely for all survey participants. The association of caffeine concentrations with blood lipids was established by means of partial correlation analysis and regression analysis. After controlling for influence factors, caffeine concentrations were closely positively related to triglycerides in caffeine-drug users (male: r = 0.245, p < 0.001; female: r = 0.117, p = 0.018) and related to HDL-C in female nonusers (r = 0.245, p < 0.001). No associations were found between caffeine concentrations and total cholesterol and LDL-C levels in any groups of our study. Acute intake of caffeine might increase triglycerides levels especially in men while chronic intake of caffeine might be weakly but positively associated with HDL-C concentrations in women. The effects of caffeine on the pathogenesis of cardiovascular diseases should be further confirmed through epidemiological studies.