BACKGROUND: Sunscreens may allow overexposure to ultraviolet A (UVA) in fair-skinned persons and prevent symptoms of sunburn, but their benefits for the prevention of melanoma are uncertain. METHODS: A PubMed search was performed that identified all known studies of the association of sunscreen use with melanoma risk during 1966-2007. A total of 18 studies were identified, of which 17 met criteria for inclusion in the analysis. Of these, 10 were conducted at latitudes >40 degrees from the equator and 7 at <or=40 degrees . Data were pooled for all latitudes combined and also according to these latitude strata. The association of skin pigmentation and latitude with odds ratios was estimated using linear regression. RESULTS: Overall, there was no statistically significant effect of use of sunscreens on risk of melanoma (odds ratio 1.2, 95% confidence interval [95% CI] 0.9-1.6; p for heterogeneity < 0.0001). However, there was an interaction with latitude. At >40 degrees from the equator, the odds ratio was 1.6 (95% C.I. 1.3-1.9; p for heterogeneity = 0.006), whereas it was 0.7 at <or=40 degrees (95% C.I. 0.4-1.0; p for heterogeneity = 0.0002). CONCLUSIONS: Use of common sunscreen formulations that absorb UVB almost completely, but transmit large quantities of UVA, may contribute to risk of melanoma in populations at latitudes >40 degrees.