To support a mixture risk assessment with a focus on male reproductive health, we conducted a systematic review of associations between bisphenol A (BPA) exposures and declines in semen quality, based on animal and epidemiological studies. Contrary to a widely held view that there is "conflicting" evidence of such associations, our review and confidence rating approach reveals that animal studies provide convincing evidence of declines of semen quality after gestational BPA exposures. Many of the reported negative findings can be attributed to deficiencies in study sensitivity, insufficient control of background contamination and probable confounding through hormonal interference due to the use of soy-containing diets. We did not evaluate animal studies of adult BPA exposures. Divergent findings in "medium to high" and "medium" confidence epidemiological studies can be explained in terms of differences in exposure conditions. We attempted the estimation of a BPA reference dose based on animal studies. Due to variations in the no-observed adverse effect levels (NOAELs) in high confidence studies, possible reference doses ranged from 0.0001 to 0.0099μg/kg/d. In choosing 0.003μg/kg/d we struck a balance between caution suggested by studies at the lower end of the doses and the weight of evidence from studies with higher NOAELs. This weighting was motivated by the intended use of the value in a mixture risk assessment which meant arriving at a reasonable estimate of BPA exposures likely without effects on semen quality. We realise that our approach does not conform with the standards necessary for deriving tolerable daily intakes (TDIs) for single chemical exposures, which is not our interest here. BPA exposures currently experienced by European populations and beyond are in excess of 0.003μg/kg/d and even fall in the range where some epidemiological studies observed effects on semen quality as a result of BPA exposures in adulthood.