Risk prediction models are useful in clinical decision making. We have published an internally validated prediction tool for lung cancer based on easily obtainable epidemiologic and clinical data. Because the precision of the model was modest, we now estimate the improvement obtained by adding two markers of DNA repair capacity. Assay data (host-cell reactivation and mutagen sensitivity) were available for 725 White lung cancer cases and 615 controls, all former or current smokers, a subset of cases and controls from the previous analysis. Multivariable models were constructed from the original variables with addition of the biomarkers separately and together. Pairwise comparisons of the area under the receiver operating characteristic curves (AUC) and 3-fold cross-validations were done. For former smokers, the AUC and 95% confidence intervals were 0.67 (0.63-0.71) for the baseline model and 0.70 (0.66-0.74) for the expanded model. For current smokers, the comparable AUC values were 0.68 (0.64-0.72) and 0.73 (0.69-0.77). For both groups, the expanded models were statistically significantly better than the baseline models (P = 0.006 and P = 0.0048, respectively), although the increases in the concordance statistics were modest. We also recomputed 1-year absolute risks of lung cancer as described previously for two different risk profiles and showed that individuals who exhibited poor repair capacity or heightened mutagen sensitivity had increased absolute risks of lung cancer. Addition of biomarker assays improved the sensitivity of the expanded models.