BackgroundBisphenol A (BPA) and, with increasing occurrence, its analogs bisphenol S (BPS) and bisphenol F (BPF) are applied in many consumer products, leading to humans being exposed from a vast number of sources and via several routes. Estrogenic and anti-androgenic effects are exerted by the chemical BPA, and also by its analogs. Therefore, realistic exposure assessments are needed for assessing risks related to cumulative exposure. ObjectivesBiomonitoring for BPA, BPS, and BPF was conducted in a human study embedded in the EU project EuroMix and the measured urinary concentrations were compared to source-to-dose calculations for source allocation and plausibility test of the model. MethodsFor two 24-hour study periods separated by 2–3 weeks, 144 adult volunteers in Norway kept detailed diaries on food consumption, personal care product (PCP) use, and thermal paper (TP) handling. Concurrently, 24 h urine was collected and urinary levels of BPA, BPS, and BPF were analyzed using ultra-high performance liquid chromatography and tandem mass spectrometry (UPLC-MS-MS). In line with the information obtained from the first study day, bisphenol exposure from food, PCPs, TP, and dust was modeled primarily individual-based with probabilistic models. Estimates for BP excretion over 24 h were obtained with the models and compared to measured amounts. ResultsModeled aggregate internal exposures covered the full range of measured urinary amounts for all BP analogs. In general, individual-based medians of modeled BPA exposures were in good agreement with the measurements, but individual-specific correlation was lacking. Modeled exposures mostly underestimated BPS and BPF levels in participants with positive measurements (53% and 8%), except for the P95 values of modeled BPS exposure that were higher than measured amounts if TP was handled. Most likely, diet and TP were the sources contributing the most to BP exposure in this study. Urinary measurements did not reveal a significant correlation between the amounts of canned food consumed, the number of PCPs used, or the number of TP handling events and levels of BPA, BPS, or BPF. ConclusionsThe good agreement between the ranges of modeled BPA exposure and measured BPA amounts indicates that available concentrations, especially from the main exposure source food, mirror the exposure situation realistically, and suggests that the exposure model considers the relevant exposure sources. The lack of individual-specific correlations means that the individual measured amounts and modeled exposures did not vary in parallel, e.g. due to mismatch of BP concentrations in food, TP, and other sources, or delayed internal exposure. The underestimation of modeled BPS and BPF exposure suggests that not all relevant sources were included in the respective exposure models. This could be due to a lack of input data, e.g. for food items, or due to an increased replacement of BPA with structural analogs compared to the used concentration and occurrence data.