SummaryIn the course of an investigation of melanocytic naevus development in Queensland, Australia, whole-body naevus counts of 66 adolescents were performed separately by two nurse examiners on two occasions on average 4 weeks apart. There was good agreement between the two examiners for counts of total naevi on the whole body (intra-class correlation coefficient = 0.96) and at selected subsites (face, neck, back, upper arms, lower arms). Agreement was lower when raised naevi only were counted (0.83). Intra-examiner repeatability was high for both nurses, particularly for the more experienced examiner (intra-class correlation coefficients = 0.98 and 0.91 for total naevi on the whole body), and was consistently better when all naevi were counted rather than naevi of a particular size. Independent counts of naevi on the back using a computer imaging technique were reproducible (intra-class correlation coefficient = 0.92), but showed only moderate agreement with counts by the nurse examiners. Overall, these results demonstrate high comparability of naevus counts between and within similarly trained examiners. They do not support the common practice in epidemiological studies of restricting counts to naevi larger than 2 mm, or of counting raised naevi only.Melanoma is an increasingly important public health problem in white-skinned populations throughout the world. In Australia, it is currently the second most common cause of cancer death, and its rapidly increasing incidence is expected to overtake that of all other cancers, with the exception of non-melanoma skin cancer, within the next decade (MacLennan et al., 1992). The number of naevi on the body is the strongest known predictor of melanoma risk and, as such, the reliable and accurate measurement of naevus numbers is important for ongoing individual risk assessment, in studies of naevus prevalence between and within populations over time and in aetiological investigations, in which the number of naevi is invariably a potential confounder of other melanoma risk factors (Green & Swerdlow, 1989).Counting naevi and consistently differentiating these from freckles and other pigmented lesions is one of the more difficult tasks in melanoma research (Green et al., 1991). Nevertheless, some limited data available indicate that reasonable agreement of naevus counts between and within examiners is possible. Roush et al. (1991) reported good agreement between examiners' counts of all naevi on the whole body (intra-class correlation coefficient = 0.92) and arm (0.88) of 153 melanoma patients at the Yale Melanoma Unit. Agreement for raised naevi on the arm was considerably lower (0.36) because of differences of opinion about what was considered palpable. In a small study in Canada (Walter et al., 1991), intra-examiner repeatability of naevus counts on the mid-left arm of eight volunteers, measured by the intra-class correlation coefficient, varied from 0.55 to 0.81 for the five examiners, and tended to be highest among those most experienced.Both of these studies were con...