Using the human XPD (ERCC2) gene as an example, we evaluate the suggestion that polymorphisms in DNA repair genes lead to decreased DNA repair capacity and to increased cancer susceptibility. This intuitively appealing idea provides the rationale for a large number of studies that have attracted much attention from scientists, clinicians and the general public. Unfortunately, most of this work presupposes that a functional effect has been established for the DNA repair gene polymorphisms under study. For XPD, there is no credible evidence for any effect on DNA repair of the two common polymorphisms leading to p.D312N and p.K751Q amino acid variations, and evolutionary analyses strongly predict that both polymorphisms are benign. Current evidence suggests no causal relationship between XPD polymorphisms, reduced DNA repair and increased cancer risk. We do not believe that more studies of the same kind will be useful. Instead, we suggest a combination of several other approaches, which up to now have been used in only a sporadic way, to examine more rigorously the possibility that phenotypic differences are associated with polymorphisms in other DNA repair genes.