Partitioning the evapotanspiration (ET) flux in a forest into its component fluxes is important for understanding the water and carbon budgets of the ecosystem. We use non-linear parameter estimation to determine the vertical profile of the Lagrangian timescale (T L) and partitioning of ET that simultaneously optimise agreement between modelled and measured vertical profiles of temperature, water vapour, carbon dioxide concentrations, and deuterated water vapour for a two-week period in November 2006. High precision real-time trace gas measurements were obtained by FTIR spectroscopy. Modelled temperature and concentration profiles are generated using a Lagrangian dispersion theory combined with source/sink distributions of HDO, H 2O, sensible heat, and CO 2. These distributions are derived from an isotopically enabled multilayer Soil Vegetation Atmospheric Transfer (SVAT) model subject to multiple constraints. The soil component of the model was tested in isolation using measured deuterium content of soil chamber evaporate, while the leaf component was tested using isotopic analyses of leaf and xylem water, combined with leaf-level gas exchange measurements. Optimisation of T L and the partition of ET was performed twice: once using only temperature, H 2O and CO 2 profiles and a second time including HDO as well. The modelled vertical concentration profiles resulting from inclusion of HDO in the cost function demonstrate our ability to make consistent estimates of both the scalar source distributions and the deuterium content of the water vapour sources. However, introducing measurements of deuterium in water vapour does not significantly alter resulting estimates of normalised T L (0.4±0.1 at canopy top) and the partition of ET (85±2% transpiration), suggesting that the additional data and modelling required to use deuterium are not warranted for the purpose of partitioning ET using the framework presented here. © 2011.