Evidence from observational studies on light at night (LAN) exposure, sleep duration, endogenous melatonin levels, and risk for breast cancer in women is conflicting. This led us to conduct a dose–response analysis of published observational data. Pertinent studies were identified by searching Medline, Web of Science, and EMBASE through April 2013. The dose–response relationship between sleep duration, urinary 6-sulphatoxymelatonin levels, and breast cancer was assessed using the restricted cubic spline model and by multivariate random-effects metaregression. A separate meta-analysis was also carried out to calculate the relative risks (RRs) with 95% confidence intervals (CIs) for breast cancer for the comparable categories or highest levels of exposure versus the lowest levels. Twelve case–control and four cohort studies were included in the analysis. High artificial LAN exposure is associated with an increased risk for breast cancer (RR=1.17, 95% CI: 1.11–1.23), but not ambient LAN exposure (RR=0.91, 95% CI: 0.78–1.07). The summary RR for breast cancer is 1.00 (95% CI: 0.995–1.01) for an increment of 1 h of sleep per night. No significant dose–response relationship between sleep duration and breast cancer was found either for the linearity test (Ptrend=0.725) or for the nonlinearity (Ptrend=0.091) test. An increasein of 15 ng/mg creatinine in urinary 6-sulphatoxymelatonin is associated with a 14% reduced risk for breast cancer (RR=0.86, 95% CI: 0.78–0.95), with a linear dose–response trend (Ptrend=0.003). There was no evidence of substantial heterogeneity or publication bias in the analysis. Our study adds to the evidence of LAN breast cancer theory. Further research in this area is warranted.