A computer-aided search identified 16 case-control studies which speci
fically assessed sunburn as a risk factor for cutaneous malignant mela
noma. Using unadjusted estimates, a history of sunburn was associated
with significantly increased risk of melanoma in all but one study. Fo
ur studies were defined as core studies after assessment of study qual
ity; however, only two of these had sufficiently similar definitions o
f sunburn to allow pooling of results. Using pooled data, the risk of
melanoma in those ever sunburned was 2.0 (95 percent confidence interv
al [CI] = 1.6-2.6), while the highest category of sunburn exposure had
a risk of 3.7 (CI = 2.5-5.4). The suggestion that sunburns in childho
od carry greater risk of melanoma cannot be supported by pooled analys
is. This review demonstrated considerable variation in design and meth
od among the studies, and identified sources of bias which prevented a
pooled analysis using all available data. The need for strong epidemi
ologic evidence relating sunburn to melanoma, particularly in childhoo
d, is of prime importance, since avoidance of sunburn is one of the fe
w potential means of primary prevention of melanoma.