Establishing mixed-species stands is frequently proposed as a strategy to adapt forests to the increasing risk of water scarcity, yet contrasted results have been reported regarding mixing effects on tree drought exposure. To investigate the drivers behind the spatial and temporal variation in water-related mixing effects, we analysed the δ13C variation in 22-year tree ring chronologies for beech and pine trees sampled from 17 pure and mixed pine–beech stands across a large gradient of environmental conditions throughout Europe. In the pure stands, average δ13C values were lower for beech (−27.9‰ to −22.2‰) than for pine (−26.0‰ to −21.1‰), irrespective of site conditions. Decreasing SPEI values (calculated over June to September) were associated with an increase in δ13C for both species, but their effect was influenced by stand basal area for pine and site water availability for beech. Mixing did not change the temporal constancy of δ13C nor the tree reaction to a drought event, for any of the species. While the mixing effect (Δ δ13C = δ13C pure stands − δ13C mixed stands) was on average positive for beech and non-significant for pine across the whole gradient, this effect strongly differed between sites. For both species, mixing was not significant at extremely dry sites and positive at dry sites; on moderately wet sites, mixing was positive for beech and negative for pine; at sites with permanent water supply, no general patterns emerge for any of the species. The pattern of mixing effect along the gradient of water availability was not linear but showed threshold points, highlighting the need to investigate such relation for other combinations of tree species.