Irrigation and fertilization both affect the water cycle in agricultural ecosystems. It is difficult to quantify root water uptake (RWU) which varies with crop development and seasons. In this study, a Bayesian mixing model (MixSIAR) coupling with dual stable isotopes (D and 18 O) was used to quantify RWU patterns for winter wheat under different irrigation and fertilization treatments between 2014 and 2015 in Beijing, China. The main RWU depth during the greening-jointing, jointing-heading, heading-filling, and filling-harvest stages was 0-20 cm, 20-70 cm, 0-20 cm, and 20-70 cm, respectively, which showed water uptake proportions of 67.0%, 42.0%, 38.7%, and 34.9%, respectively. Significant differences in RWU patterns appeared between the 2014 and 2015 seasons. The main RWU depth increased gradually from 0-20 cm at the greening-jointing stage to 20-70 cm at the jointing-heading stage and 70-150 cm during the heading to harvest period in 2014. However, winter wheat primarily took up soil water from the 0-70 cm layer in 2015. The average water uptake proportion in the top layer (0-20 cm) in 2015 (42.6%) was remarkably higher than that in 2014 (28.7%). There was a significantly negative relationship (p < 0.01) between the water uptake proportion and the proportion of root length at the filling-harvest stage in 2014, while no significant correlation (p > 0.05) was found in 2015. Variable distributions of root characteristics and soil moisture induced by different irrigation and fertilization comprehensively affected the RWU profile, particularly under severe drought environments in 2015. Treatments with fertilization of 105 kg hm −2 N or irrigation of 20 mm during the greening-jointing stage significantly promoted water uptake contribution in the 70-150 cm (32.2%) and 150-200 cm (23.5%) layers at the jointing-heading stage in 2015, while other treatments had a shallow dominant RWU depth (0-20 cm). The planned wetting layer should be kept within the main RWU depth of 0-70 cm for improving irrigation water use efficiency.Since RWU generally does not lead to isotopic fractionation or the magnitude of fractionation is too low to be observable for the majority of plant species, stable isotopic analysis is a powerful and valuable tool to identify and quantify RWU via the measurements of δD and δ 18 O abundance in waters [7,8]. Widely used methods to determine RWU patterns include the IsoSource [9] and Bayesian mixing models (e.g., SIAR [10], MixSIR [11], and MixSIAR [12]). In comparison, the Bayesian mixing models are particularly efficient in the case of multiple layers for illustrating the interest of reaching the best vertical resolution and maximizing the number of identified potential sources [7]. The MixSIAR model incorporates advances in MixSIR and SIAR that considers multiple sources of uncertainty and provides potential proportions of source contributions [12]. The MixSIAR model has been employed successfully to determine the contributions of soil water at different layers to the water uptake of field crops and na...