Variation within and around the apolipoprotein C-III (APOC3) gene has been associated with elevated triglyceride (Tg) levels and cardiovascular disease. The associations of 4 polymorphic variants in the APOC3 gene (3238C>G in the 3' untranslated region [SST:I], 1100C>T in exon 3, -482C>T in the insulin-responsive element, and -2854T>G in the APOC3-A4 intergenic region) with plasma Tg and cholesterol levels and their interaction with smoking have been investigated in the Second Northwick Park Heart Study (NPHSII), a large cohort of healthy men (n=2745). Analyzing the variants separately showed that 3238G, 1100T, and -482T alleles were all associated with raised Tg levels. For the 3238C>G and -482C>T sites, the Tg-raising effect appeared to depend on smoking status (test for interaction, P:=0.042 and P:=0.009, respectively), but for the 1100C>T site, the effect was constant irrespective of smoking status (test for interaction, P:=0.27). The -2854T>G site was not associated with effects on Tg levels in this sample. Because all of the variants showed significant allelic association, regression modeling was used to quantify the relative size of each effect and to assess whether the effects of the separate variants were independent. The 1100C>T variant had an independent effect on Tg levels that was not influenced by smoking status (increase of 8.2% in Tg with each T1100 allele), whereas the -482C>T variant had a separate effect that was dependent on smoking (increase of 13.7% in Tg for each -482T allele in current smokers, 8.6% in exsmokers, and -7.4% in those who never smoked). The 3238C>G variant did not show a separate independent effect on Tg concentration. Thus, by use of the regression model, it was possible to estimate how mean Tg levels would vary in groups of individuals with respect to APOC3 genotype and smoking information. Analysis in this large group of healthy men has allowed the identification of a statistically robust APOC3 genotype-smoking interaction, which now warrants further molecular study.