The authors implemented the Regional Ocean Modelling System (ROMS) to coastal waters on the west coast of Ireland. The Connemara model has c.200 m horizontal resolution and has 20 vertical sigma levels and stretches from 10.8oW to 8.9oW and from 52.95oN to 53.73oN. It has three open ocean boundary conditions in the north, south and west, and several rivers are included at the head of Galway Bay. Both 3D and 2D configurations have been set up; with the aim of the 2D configuration being the prediction of storm surges. The 3D configuration has been running in the operational forecasting mode for approximately the last decade and produce weekly hindcasts and 3-day forecasts. Details of the operational system will be presented. Most recent developments include the implementation of wetting/drying algorithm with a critical depth of 0.25 m. It has undergone a testing phase and the authors will report on the findings in terms of the computational efficiency and in terms of the changes to the model skill. Details of the set-up of each configuration, as regards the forcing functions, the choice of boundary conditions, atmospheric forcing, advection schemes, turbulence schemes, will be presented. The predictive skill of various configurations have been assessed by means of root mean square error (RMSE) differences and correlations with data collected from the Marine Institute’s observational network in Galway Bay. The observational platforms comprise of 5 tide gauges, 4 x ADCPs and temperature and salinity sensors. The harmonic analysis has been carried out on the data from the tide gauges and the corresponding model predictions in order to validate the tidal signal. The tidal signal in the Sea Surface Height (SSH) data is dominated by three semi-diurnal constituents (i.e. constituents with a period of approx. 12 to 12.5 hours), M2, S2 and N2, and three diurnal constituents, K1, O1 and Q1. At Galway Port tide gauge, the M2 magnitude error between the model and data is 0.07 m. The surge component is calculated for observed and modelled data at the locations of tide gauges. The authors will present the skill of the model as regards the prediction of storm surges to include the comparison between a 3D and 2D configuration.