The production of synthetic fertilisers from traditional sources has several issues regarding sustainability, particularly the energy intensity required for production, limited mineral resources and loss of terrestrial environment. Nutrient recovery from wastewaters is an opportunity for alternative fertiliser manufacture, largely due to the high quantities of nutrients present. Electrodialysis (ED) is an optimal technology to recover, particularly, NH4+ -N and K+-K. In this thesis, a mechanistic modelling approach is used to study ED for nutrient recovery. Existing ED models lack the capacity to mechanistically evaluate complex, multi-ion solutions (like real wastewater), the integration of electrochemistry and physicochemical phenomena across the whole reactor domain, and model validation using system dynamics. These limitations are addressed by development of a mechanistic modelling approach, with key mechanisms determined through targeted laboratory scale experiments using synthetic and real wastewaters. The first major study in this thesis includes development of a mechanistic model validated using dynamic experiments fed with synthetic wastewater. It was found that membrane resistance was the major contributor to potential drop and that apparent boundary layers were relatively thick (3 ± 1 mm) due to reactor design and operational conditions. This study also established that non-ideal solution effects such as ion pairing and ionic activity had a major impact, enhancing the capability of ED to recover monovalent nutrient ions such as K+ and NH4+. The second major study used real centrifuged anaerobic digester rejection water (centrate) to study membrane scaling. Dynamic laboratory experiments demonstrated electro-concentration of nutrients in centrate to several times the feed concentration. An 87±7% by weight reduction in scale occurred when the centrate was pre-treated by upstream struvite recovery and the product pH was controlled at pH 5, compared to untreated centrate and no pH control. A mechanistic model for the inorganic processes was developed by extending the previously developed model to include the composition of a real wastewater feed solution and precipitation. This extended model revealed a reduction in struvite scale to the removal of phosphate during the struvite pretreatment, and reduction in calcium carbonate scale to pH control resulting in the stripping of carbonate as carbon dioxide gas; indicating that multiple strategies may be required to control precipitation. A third major study evaluated the mechanisms limiting high product concentrations during electro-concentration of synthetic urine. Modelling this system using similar methods to the first study identified that high concentrations in the product are prevented by back diffusion of ions across the membrane, current leakage (when buffering capacity is exhausted), and water fluxes across the membranes. Mechanistic discoveries in this thesis provide practical guidelines for pilot and full-scale operation of nutrient electroconcentration. The process of model development has extended the existing literature to quantify previously difficult to measure mechanisms such as competitive current transport between multiple ionic species, current leakage related to pH and solution composition, mechanisms for precipitation, and limitations to high product concentrations. Modelling methods developed here aim to be nonspecific and may be readily adapted to study other electrochemical membrane systems, thereby encouraging future applications.