There is growing evidence that individuals at high risk of cutaneous melanoma can be identified by the use of simple measures of benign melanocytic naevi--raised risk occurring in persons who have large numbers of naevi, or naevi with atypical clinical features. Very high risk of melanoma appears to exist in two rare groups: persons with dysplastic naevi whose families include at least two individuals who have had melanoma, and persons with giant congenital naevi. Risks of melanoma in other individuals with dysplastic naevi and in persons with small congenital naevi have not yet been quantified. Much of the published work on risk of melanoma in relation to naevi has been based on clinical series or assemblage of case reports, with great potential for bias. In such studies, naevi have often been classified into dichotomies or syndromes originally defined for pathological or clinical purposes, frequently using poorly reproducible criteria. Case-control and cohort studies of the relationship of naevi to melanoma are needed, which use reproducible criteria and designs minimizing bias, and which examine risk in relation to graded clinical and histological measures of naevi.