Despite strong evidence that sun exposure causes malignant melanoma, the details of this relation remain unclear. A nested case-control analysis was conducted within the Nurses' Health Study cohort to examine the relation between timing of severe sun exposure and incidence of melanoma. The subjects were 130 white women aged 38 to 65 years with confirmed cutaneous melanoma (other than acral lentigenous) who reported no history of melanoma in first-degree relatives. The control subjects were 300 women matched by race, date of birth, and cycle of questionnaire who also reported no history of melanoma in first-degree relatives. We used conditional logistic regression to evaluate the relation of sun damage after 30 years of age and sun damage from 15 to 20 years of age to the incidence of melanoma. Blistering sunburns between 15 to 20 years of age were associated with risk of melanoma (relative risk = 2.2 for five or more burns vs none, 95% confidence interval 1.2 to 3.8). This association persisted when a history of burns after 30 years of age was controlled in the analysis. No material association was found between blistering sunburns after 30 years of age and melanoma. Similarly, a more equatorial latitude of residence between 15 and 20 years of age was positively associated with melanoma; latitude after 30 years of age was less strongly and not significantly related to melanoma risk. Sun exposure prior to 20 years of age is more closely associated with melanoma risk than sun exposure after 30 years of age.