For New Zealand, a country straddling the Pacific ‘Ring of Fire’, effective mitigation of the risks posed by tsunamis is an urgent priority. Mitigation measures include evacuation mapping, land-use planning and engineering of tsunami resilient buildings and infrastructure; but for these to be effective, a quantitative estimate of the tsunami hazard is needed. For this purpose we present the New Zealand Probabilistic Tsunami Hazard Model (NZPTHM). The model uses a Monte Carlo method for sampling from the geophysical parameters that constrain the magnitude–frequency distributions of the earthquake sources that can cause tsunamis affecting New Zealand. The sampled parameters are used to construct synthetic catalogues of the source events and the subsequent tsunami heights. Processing of these synthetic catalogues produces hazard curves, describing maximum tsunami height as a function of return period, which include ‘error bars’ (confidence intervals) as determined by the Monte Carlo model. Most practical mitigation measures require inundation modelling, and for this purpose we propose using de-aggregation, a process by which a small set of scenarios can be extracted from the NZPTHM for the purpose of detailed inundation modelling.