In this work, precipitable water vapour (PWV), its diurnal variability and water vapour vertical profiles have been obtained from Weather Research and Forecasting (WRF) model simulations. The PWV simulated values have been validated using observations from the Global Navigation Satellite Systems (GNSS) for three locations in an orographically complex region, the Canary Islands archipelago, while the vertical water vapour profiles have been compared with radiosonde data available for this region. Initial and boundary conditions for simulations are given by National Center for Environmental Prediction (NCEP) global models: FNL (Global Final Analysis) and GFS (Global Forecast System). Several parametrizations and assumptions, related to cloud microphysics, boundary-layer and land surface models, number of vertical levels, moisture transport and 3D nudging techniques, have been applied to evaluate the sensitivity of the WRF results. In general, a good agreement between WRF and GNSS data was obtained, with correlation coefficients larger than 0.95 and centred root mean square differences about 2 mm. Furthermore, for PWV and water vapour vertical distribution, no appreciable differences were observed between the different experiments carried out, since the main contribution to water vapour in a region with small land coverage comes from the large-scale processes, which are reasonably represented when analysis data are used as the forcing boundary conditions of the model. The ability of the WRF simulations to reproduce the PWV diurnal cycle was also verified. Additionally, the skills of the WRF simulations to determine PWV were evaluated for different nudging strategies when 48 h forecast simulations were performed; these showed insignificant differences.