A study of all newly incident melanoma patients in British Columbia in 1991-1992 was undertaken to test the hypothesis raised by an earlier study, which showed that in younger patients the incidence rate of melanoma per unit area of skin was higher on intermittently exposed skin areas than on continuously exposed areas. Using 1,033 patients and a more detailed body site categorisation than was previously possible, our results confirmed that in both men and women under age 50 the highest melanoma density was on the back. At ages over 50, the greatest density occurred on fully exposed sites, such as the face, though the dorsum of the hand and forearm, likely also to have high exposure, show very low melanoma densities. Differences between males and females correlate well with differences in likely exposure patterns. These results were seen for all invasive cutaneous melanomas combined; the patterns were similar for subtypes and for both invasive and in situ melanoma, with the exception of lentigo maligna melanoma (LMM), which occurs almost exclusively on the face, even at younger ages. Comparison with the earlier study (1976-1979) shows that the age-standardised rates for melanoma excluding LMM have increased by 60%, with the greatest proportional increase being at younger ages; in the recent data, the age-standardised rate for intermittently exposed sites exceeds that for usually exposed sites. Our results confirm that intermittent sun exposure has a greater potential for producing melanoma than continuous exposure at ages below about 50, though at older ages melanoma is more common on body sites with continuous sun exposure.