Incident cases of in situ and invasive cutaneous malignant melanoma diagnosed during 1975-90 were identified through the National Cancer Institute's Surveillance, Epidemiology, and End Results program. We studied the 32 868 white subjects diagnosed with melanoma, who were living in nine cancer registry areas covering approximately 10% of the population of the USA. The summer-to-winter ratio, defined as the ratio of the number of melanomas diagnosed during June to August (summer), to the number of melanomas diagnosed during December to February (winter), was determined according to gender, stage, histologic type and anatomic site. Summer-towinter ratios were 1.47 (95% confidence interval (Cl) 1.37−1.58) for in situ; 1.43 (95% Cl 1.38−1.48) for local stage; 1.24 (95% Cl 1.12−1.38) for regional stage; and 0.95 (95% Cl 0.82-1.11) for distant stage melanoma. For the melanomas staged as local at diagnosis (86% of the invasive melanomas staged), a July peak was observed. For each of the major histological types of local stage melanoma, summer-to-winter ratios were significantly elevated in men (range 1.24−1.41) and women (range 1.44−1.90). For the major anatomic sites (including the head and neck, which are exposed throughout the year) of local stage melanoma, summer-to-winter ratios were elevated for men (range 1.28−1.45) and for women (range 1.31−1.75). Although some of the seasonal variation in frequency of diagnosis of cutaneous melanoma may be due to a greater likelihood of detection during the summer months, when less clothing is worn and more skin is visible, we conclude that the seasonal variation is due at least in part to relatively recent exposure to sun. These findings may hold clues to the last stages of carcinogenesis, and suggest that avoiding excessive exposure to the sun may decrease the risk of melanoma in the short term as well as the long term.