Scientists have uncovered evidence that two major earthquakes of the same type that triggered both the 2011 Japanese and 2004 Boxing Day tsunamis hit central New Zealand in the past 1000 years.
And they warn another could cause widespread damage, triggering landslides, sea level changes and devastating tsunami.
The two subduction "megathrust" earthquakes struck about 10-30km beneath the seabed in Cook Strait - one of them 520 to 470 years ago, and the other 880 to 800 years ago.
The breakthrough discovery by GNS Science had "confirmed" the risk this type of earthquake, more powerful and potentially destructive than normal quakes, poses to New Zealand.
Both the 2011 Japanese tsunami and 2004 Boxing Day tsunami were caused by subduction earthquakes.
GNS earthquake risk analyst Nick Horspool said if a Japanese-style quake of the same magnitude were to hit central New Zealand, thousands of people could be killed and tens of billions of dollars lost.
Mr Horspool estimated the potential impact of such an event using a computer model.
"If it was a magnitude 9, we'd have really strong shaking the whole way up the East Coast of the North Island and even the top of the South Island - intensities near [to] what was experienced in Christchurch.
"There would be widespread damage, damage to buildings and infrastructure, and following that you would then have a tsunami."
He said the tsunami would "devastate" the coastal areas and cause further destruction to buildings and infrastructure.
The exact number of casualties, he said, would depend on how many people were able to self-evacuate.
"Because it's going to be a large, strong shaking... then 10 minutes later there will be the tsunami coming."
Subduction quakes are different from normal earthquakes in that they occur on the under surface of the upper plate - where two tectonic plates meet - instead of on faults within the upper plate.
They have the potential to be bigger in magnitude than fault ruptures closer to the surface, tend to affect larger areas, and are more likely to trigger a tsunami.
The GNS study's lead author Kate Clark said the findings were significant as they helped to further understand earthquake and tsunami hazards in the lower North Island and upper South Island.
"Subduction earthquakes are not a 'new' risk for New Zealand, as we have always assumed they can occur, and they are accounted for in our seismic hazard models," said Dr Clark.
"This study is significant in that it confirms that risk."
Although the evidence from just two earthquakes was still not enough to calculate when the next subduction earthquake would occur , or its exact magnitude, she said: "It [was] most likely to be greater than magnitude 7."
It was "fairly likely" a tsunami would be generated, along with widespread coastal subsidence and sea level change, and a significant amount of landslides.
"Certainly Wellington, Napier, Hastings - the larger cities along that east coast of the North Island - and Palmerston North are likely to be affected," she said.
Masterton, Blenheim and up to Gisborne could also be hit.
Scientists have previously found evidence of subduction earthquakes occurring under Hawkes Bay, but this was the first time they had found direct geological evidence that the southern part of the Hikurangi Margin could rupture in an earthquake.
They identified the quakes from sediment cores extracted from Big Lagoon, a large coastal lake east of Blenheim.
The older of the two earthquakes identified at Big Lagoon also triggered a three-metre-high tsunami that travelled inland about 360m.
The research was outlined in the Bulletin of the Seismological Society of America, published this week.
Victoria University Associate Professor John Townend, EQC Fellow in Seismic Studies, said the study provided the first substantiated evidence for large subduction zone earthquakes in the area, and would enable New Zealand's seismic hazard model to be "refined".
"In order to understand the hazard posed to New Zealand by future large earthquakes, we need to know when such earthquakes have occurred in the past - and how big they were.
The challenge with this is that the earthquakes of most interest happen very infrequently."
