Objective: Effect modification by obesity or obesity-related phenotypes (e.g. physical activity and diet) was observed in some candidate gene studies on lipids. We aimed to evaluate gene-obesity interaction effects on HDL (HDL-C), LDL (LDL-C) and total cholesterol (TC) levels using genetic predisposition scores. Methods: We derived imputed genotypes for 104 SNPs in 84 lipid-associated loci in the population-based studies KORA F3 (n = 1406) and KORA F4 (n = 1515). We inferred specific unbiased weights for each SNP from linear regression estimates in KORA F4. Weighted genotypic predisposition SNP-scores were then calculated for each lipid trait in KORA F3. Interaction terms of SNP-scores with each of the obesity parameters (BMI, waist-hip-ratio, waist circumference) were included in age-and sex-adjusted linear regression models on HDL-C, LDL-C and TC. Results: All three SNP-scores were shown to be highly associated with their respective lipid levels (p = 1.11 x 10(-47) for HDL-C, p = 3.25 x 10(-19) for LDL-C and 1.53 x 10(-13) for TC). BMI, WHR and waist circumference modified the effect of the weighted HDL-C SNP-score on HDL-C significantly (all interaction p-values < 0.0062). No interaction term was significant for LDL-C or TC. Accounting for gene-obesity interaction substantially increased the proportion of HDL-C variance explained by 3-3.5%. Conclusion: Our investigation revealed a significant interaction effect between obesity parameters and a SNP-score on HDL-C: the combined effect of HDL-C-increasing alleles on HDL-C is attenuated with increasing levels of obesity-relevant parameters. Future studies aiming to detect new genetic variants or to model genetic predictions of HDL-C levels should take obesity or obesity associated parameters into account.