Although the Alpine Fault is "well behaved", scientists say there is a 30 percent chance of a large earthquake along the fault over the next 50 years.
A four-year study into the fault, which stretches 600km along the spine of the South Island, from Marlborough in the north to Milford Sound in the south, has given scientists a greater knowledge of how often the fault ruptures.
When it ruptures it can produce earthquakes of about magnitude eight on the Richter Scale.
The investigation, which was centred on a remote river terrace near Lake McKerrow about 35km northeast of Milford Sound, and supplemented with information from the Haast area, found evidence of 24 surface ruptures of the Alpine Fault dating back to 6000BC.
Previously scientists have only been able to date four earthquakes along the fault dating back to about 1000AD.
The last major quake on the fault occurred 295 years ago.
Scientists found the mean interval between large earthquakes on the fault is 330 years and two thirds of the intervals were between 260 and 400 years.
The longest gap without a major quake is 510 years, while the shortest is 140.
Scientists used a range of investigation techniques, including radiocarbon dating of seeds, leaves, and reeds contained in swampy sediments, to determine the ages of the ruptures.
Dr Kelvin Berryman, manager Natural Hazards Research Platform, GNS Science, said the findings were entirely new for the Alpine Fault. He said it was probably only the fifth fault in the world to have such a long record of earthquakes documented.
Dr Berryman said the findings will greatly improve the reliability of earthquake hazard modelling.
"This research has particular significance to Canterbury where the findings are being incorporated in liquefaction and earthquake shaking mitigation measures as part of the rebuilding process.
"Based on this research, scientists have estimated there is a 30 percent probability of a large earthquake on the Alpine Fault in the next 50 years. This is at the higher end of existing estimates, which are based on only the last four large quakes on the fault.
"Scientists have more confidence in this new estimate as it is based on a much longer earthquake record."
He stressed it was not a prediction of an earthquake.
"The public might misinterpret the research findings as a prediction - it is not.
"There is also a possibility of misinterpreting the results as the earthquake being imminent, which it may not be. The mean recurrence interval between the 24 earthquakes is about 330 years. So with an elapsed time of about 295 years since the last big quake, a major earthquake in the near future would not be a surprise. Equally it could be up to 100 years away.
"The bottom line is - if not in our lifetimes then increasingly likely in our children's or our grandchildren's. Therefore a precautionary approach is certainly warranted."
Dr Berryman described the fault as "well behaved", although that does not mean scientists can predict when it will rupture.
"It simply means the Alpine Fault exhibits a fairly regular cycle of stress accumulation and rupture. It does not have long periods of more than 1000 years of inactivity. Equally, it does not have clusters of big earthquakes occurring at short intervals."
"Scientists have found a fault that responds to the steady motion of tectonic plates by rupturing at reasonably regular intervals. This illustrates that at least some plate boundary faults can be 'well-behaved' when they are have high rates of movement and are not influenced by activity on other nearby faults."
Project co-leader Ursula Cochran, also of GNS Science, said the long earthquake record highlighted the inevitability of a major earthquake occurring on the southern section of the Alpine Fault in the future.
"The last earthquake occurred about 295 years ago, so there is no better time than the present to prepare for the next one," Dr Cochran said.
"The more thorough the preparation, the lower the eventual impact will be," Dr Cochran said.
The scientists involved in the research believe the models used can be applied to other similar faults around the world, such as sections of the North Anatolian Fault in Turkey and the Denali Fault in Alaska.
Professor Richard Norris, of the University of Otago's Department of Geology, said the study is of national and international significance.
"The Alpine Fault has the highest level of probability for rupture of any fault in New Zealand," Dr Norris said.
"Westland obviously is at high risk, with widespread damage likely and roads, bridges and other transport links likely to be badly affected (and the tourist trade). The fault crosses the west coast road in many places ... and with an estimated 8 m displacement, will completely destroy it.
"Intensities further east in places like Queenstown, Te Anau, Wanaka and Mt Cook will be high enough to cause landslips and do damage.
"Further east in the major cities of Christchurch and Dunedin, the intensities will be lower but the duration of shaking could still be sufficient to damage poorly constructed buildings ... and possibly cause some liquefaction.
"Places such as Nelson, Wellington and Invercargill could also expect to feel some shaking."