The “Wahine storm” was one of the two worst regenerating tropical cyclones to hit New Zealand last century. Ex-tropical systems of this type are the major source of coastal hazards in the form of severe winds and waves in the New Zealand region. At the time, the likely weather associated with the storm was well forecast, although its expected path was not. Taking advantage of recently produced reanalysis data from the National Centres for Environmental Prediction (NCEP) and using the Regional Atmospheric Modelling System (RAMS) atmospheric model and WAve Model (WAM) wave model, we have simulated the atmospheric and wave conditions associated with the Wahine storm. Using modern computing methods, the path and storm intensity were quite well forecast although uncertainties in the initial positions of the low level tropical vortex and the upper level mid-latitude trough lead to a 3-h delay in the forecast arrival of the associated southerly change at Wellington. This illustrates the need for forecasters to realise the limitations of the tools they are using. The waves forecast around the coasts of New Zealand based on the winds from this atmospheric forecast were very good between North and East Cape and down the Canterbury coast, but the model failed to produce the 10-12 m seas experienced in Wellington at the time of the sinking of the TEV Wahine. The westward error in the storm track, though a relatively modest 50-100 km, is likely to have contributed to this under-prediction. A comparison of 20 years of model hindcast waves round the New Zealand coasts with shorter periods of Waverider observations near Wellington indicated that the wave model at 1.125° resolution generally under-predicts the full degree of variability of wave height seen during extreme wave events in the Wellington region. It is likely that limited spatial and temporal resolution of the wind and wave models is a limiting factor in this respect. It was found that significant improvement in the timing and heights of the forecast waves was achieved by interpolating between the 2.5° spatially resolved and 6-h time resolved NCEP atmospheric data using a sophisticated weather prediction model. Provided the resolution of the winds driving the wave model was better than 1° and the frequency better than 1 h then large gains were made. Increasing the resolution to 0.333° increased the peak waves by a further metre but going to 0.1° made little difference except near sheltered coastal regions.