We have reviewed published case-control studies on the role of Restriction Fragment Length Polymorphism (RFLP) for specific loci in the causation of human cancer or as a prognostic factor. Five studies have been published on L-myc polymorphism and prognosis for several types of cancer (lung, kidney, stomach or breast). Sixteen studies report on Ha-ras polymorphism and the risk of cancers in the lung, breast, bladder, colorectum, brain, leukemias and melanoma. The results of the studies are conflicting. In addition, such studies raise important methodological issues: the choice of the cut-off between "rare" and "common" alleles; the inclusion of prevalent cases; the distribution of the alleles in subgroups of the population, including different ethnic groups; a mechanistic interpretation of the role played by the polymorphism, including possible interaction with environmental exposures. These questions need to be answered before a cause-effect relationship can be clearly established.