As DNA repair plays an important role in genetic susceptibility to cancer, assessment of the DNA repair phenotype is critical for molecular epidemiological studies of cancer. In this report, we compared use of the luciferase (luc) reporter gene in a host-cell reactivation (HCR) (LUC) assay of repair of ultraviolet (UV) damage to DNA to use of the chloramphenicol (cat) gene-based HCR (CAT) assay we used previously for case-control studies. We performed both the assays on cryopreserved lymphocytes from 102 healthy non-Hispanic white subjects. There was a close correlation between DNA repair capacity (DRC) as measured by the LUC and CAT assays. Although these two assays had similar variation, the LUC assay was faster and more sensitive. We also analyzed the relationship between DRC and the subjects' previously determined genotypes for four polymorphisms of two nucleotide-excision repair (NER) genes (in intron 9 of xeroderma pigmentosum (XP) C and exons 6, 10 and 23 of XPD) and one polymorphism of a base-excision repair gene in exon 10 of X-ray complementing group 1 (XRCC1). The DRC was significantly lower in subjects homozygous for one or more polymorphisms of the two NER genes than in subjects with other genotypes (P=0.010). In contrast, the polymorphic XRCC1 allele had no significant effect on DRC. These results suggest that the post-UV LUC assay measures NER phenotype and that polymorphisms of XPC and XPD genes modulate DRC. For population studies of the DNA repair phenotype, many samples need to be evaluated, and so the LUC assay has several advantages over the CAT assay: the LUC assay was more sensitive, had less variation, was not radioactive, was easier to perform, and required fewer cryopreserved cells. These features make the LUC-based HCR assay suitable for molecular epidemiological studies.