<p>Plant water source tracing studies often rely on differences in stable isotope composition of different water sources. However, an increasing number of studies has indicated a discrepancy between the isotopic signature of plant xylem water and the water sources assumed to be used by plants. Based on a meta-analysis we have reconfirmed this discrepancy between plant xylem water and groundwater and suggest back-flow of leaf water (BFLW), defined as a combination of (i) the Péclet effect, (ii) foliar water uptake (FWU) and (iii) hydraulic redistribution of leaf water, as a possible explanation for these observations. Using the average 2.21 ‰ <sup>18</sup>O enrichment of xylem water compared to groundwater in our meta-analysis, we modelled the potential of BFLW to result in this observed isotopic discrepancy. With a low flow velocity of 0.052 m.h<sup>−1</sup> and an effective path length of 2 m, the Péclet effect alone was able to account for the average offset between xylem water and groundwater. When including a realistic fraction of 5–10 % xylem water originating from FWU and tissue dehydration, 60–100 % of the average observed enrichment can be explained. By combining the Péclet effect with FWU and tissue dehydration, some of the more extreme offsets in our meta-analysis can be elucidated. These large effects are more probable during dry conditions when drought stress lowers transpiration rates, leading to a larger Péclet effect, more tissue dehydration, and a potential greater contribution of FWU.</p>