Stable carbon isotopes in plants can help evaluate and improve the representation of carbon and water cycles in land-surface models, increasing confidence in projections of vegetation response to climate change. Here, we evaluated the predictive skills of the Joint UK Land Environmental Simulator (JULES) to capture spatio-temporal variations in carbon isotope discrimination (Δ13C) reconstructed by tree rings at 12 sites in the United Kingdom over the period 1979–2016. Modeled and measured Δ13C time series were compared at each site and their relationships with local climate investigated. Modeled Δ13C time series were significantly correlated (p < 0.05) with tree-ring Δ13C at eight sites, but JULES underestimated mean Δ13C values at all sites, by up to 2.6‰. Differences in mean Δ13C may result from post-photosynthetic isotopic fractionations that were not considered in JULES. Inter-annual variability in Δ13C was also underestimated by JULES at all sites. While modeled Δ13C typically increased over time across the UK, tree-ring Δ13C values increased only at five sites located in the northern regions but decreased at the southern-most sites. Considering all sites together, JULES captured the overall influence of environmental drivers on Δ13C but failed to capture the direction of change in Δ13C caused by air temperature, atmospheric CO2 and vapor pressure deficit at some sites. Results indicate that the representation of carbon-water coupling in JULES could be improved to reproduce both the trend and magnitude of interannual variability in isotopic records, the influence of local climate on Δ13C, and to reduce uncertainties in predicting vegetation-environment interactions.