Background It is hypothesised that sleep duration affects the production of melatonin and subsequently influences cancer risk. The relationship between sleep duration and risk of breast cancer has been investigated recently but findings from epidemiological studies have been inconsistent and have not been summarised quantitatively. We aimed to conduct a meta-analysis of observational studies of the association between sleep duration and breast cancer. Methods Relevant publications were identified from reviews and computer-aided searches using PubMed, with keywords “sleep duration”, “breast cancer”, “survival rate”, “mortality”, “morbidity”, “incidence” and “risk”, up to July 23rd, 2009. RR estimates and 95% CIs were extracted for the comparison between the highest exposure group, women who slept ≥ 9 hours, and the reference group that comprised women who had a moderate sleep duration (7 or 8 hours). Summary RRs were estimated by calculating the average of the log RRs, weighted by the inverse variances of the log RRs. Results Five studies, four with prospective data and one case-control study, were identified on the risk of breast cancer in relation to sleep duration. The published data include 9166 women with incident invasive breast cancer and 147 344 women without breast cancer. When results from these studies were combined, the aggregate RR was 0.96 (95% CI 0.86 to −1.07) for women with the longest sleep duration compared to those in the reference group with a shorter sleep duration. When analysis was restricted to prospective data, the aggregate RR was 0.89 (95% CI 0.78 to 1.01). There was no evidence for significant heterogeneity in this association by menopausal status. Conclusion Meta-analysis of the published epidemiological data provides no strong evidence for a relationship between sleep duration and risk of breast cancer.