Shelterbelts are used for a variety of purposes in agricultural environments, primarily because of their ability to improve the downwind microclimate. Excessive evaporative losses from small, agricultural water supply reservoirs in semi-arid Western Australia motivated a combined numerical modelling and field investigation into the potential for using shelterbelts to reduce evaporation from these open waterbodies. A numerical model of the disturbed momentum and turbulence fields in the region modified by the wind-shelter was employed and accounted for the presence of a waterbody downwind. The model was coupled with conservation equations for heat and moisture and sensible and latent heat fluxes were estimated from the simulated momentum, temperature and humidity fields. The numerical simulations were tested against four days of field data from two experiments conducted in the agricultural districts of southwest Western Australia that measured boundary-layer evolution over a variety of small waterbodies protected by artifical and natural wind-shelters. The model provided good predictions of windspeed during neutral conditions, but inadequate specification of the upwind boundary during non-neutral stabilities resulted in the model failing to capture any sensitivity to atmospheric stability as seen in the field data. Despite this limitation, the temperature and humidity fields were adequately captured by the model, and evaporative mass flux predictions also agreed well with estimates taken from water-balance measurements. It is concluded that well-designed wind-shelters can reduce evaporation from open waterbodies by 20-30% as a result of reductions in the velocity scales responsible for removing moisture from the water surface. The model can be used to estimate the values of various shelterbelt design parameters (e.g., porosity, height) that could be applied in the field to provide optimum evaporation reductions.