Leaf intrinsic water use efficiency (iWUE) quantifies the trade-off between carbon assimilation and water loss in plants, and is constrained by leaf traits such as maximum carboxylation capacity (Vc,max) and stomatal conductance. Yet, the potential links of iWUE with leaf elementomes across different forest types remain unclear. Here, we analyzed iWUE (estimated by leaf carbon isotopes) variability and its associations with Vc,max, stomatal conductance (estimated by 18O enrichment in leaf dry matter above source water, Δ18O), and leaf elementomes across 82 tree species from temperate, subtropical and tropical forests, and evaluated the effectiveness of leaf reflectance spectroscopy as an indicator of iWUE variability and trait–iWUE associations. Across species, Vc,max, Δ18O, leaf mass per area (LMA) and leaf iron, nitrogen, sodium and manganese concentrations were the traits most strongly associated with cross-site iWUE variability. Furthermore, climatic factors (mean annual precipitation, mean annual temperature and climate moisture index) shaped trait–iWUE covariation by negatively linking leaf elements and positively with LMA, which affected iWUE more directly than indirectly via Vc,max and Δ18O. Leaf reflectance spectroscopy accurately predicted iWUE (R2 = 0.83), and the trait–iWUE relationships derived from spectral modelling were consistent with those obtained through field measurements. These findings reveal strong linkages between the leaf elementomes and iWUE, and highlight the potential of reflectance spectroscopy for characterizing iWUE variability and trait–iWUE relationships, thereby improving process modelling of forest carbon and water cycles.