The eruption of the Lusi mud volcano in Indonesia was caused by drilling for oil and gas, a meeting of 74 leading geologists has concluded.
Lusi erupted in May 2006 and continues to spew out boiling mud, displacing around 30,000 people in East Java.
Drilling firm Lapindo Brantas denies a nearby well was the trigger, blaming an earthquake 280km (174 miles) away.
Around 10,000 families who have lost their homes are awaiting compensation, which could run as high as $70m (£43m).
This is the data we wanted to get out - the data I have never been able to show before. It clearly shows that the well failed. It was the driver for the eruption
Professor Richard Davies
After debating new evidence at a conference in South Africa, most geologists voted drilling as the cause.
Correspondents describe the result a significant development in the tug-of-war to establish liability for the disaster.
The debate on the cause of the eruption took place at a meeting of the American Association of Petroleum Geologists, in Cape Town.
It was the first time the two opposing sides had agreed to debate before an international conference of independent experts.
The contest was chaired by a professional football referee - Professor John Underhill, an Edinburgh University geologist, who is also a match official in the Scottish Premier League.
We presented clear and indisputable facts that none of the four required factors for the well to have been responsible for triggering the eruption occurred.
Lapindo Brantas spokesman
The dispute centres on some newly released data - measurements taken from the Banjar-Panji-1 exploration well during the final 24 hours leading up to the eruption.
Professor Richard Davies, of Durham University in the UK, argued that these readings clearly point to a build up of pressure, causing fractures which propagated from the bore hole to the surface 150m away, resulting in the eruption.
However, Rocky Sawolo, senior drilling adviser of Lapindo Brantas, used the same primary data to argue the opposite - the pressure within the well was within acceptable limits.
Satellite photo showing the devastation caused by the Lusi mud volcano
His colleague Dr Adriano Mazzini, of the University of Oslo, testified that the fracture was triggered by a magnitude 6.3 earthquake two days earlier, centred on Yogyakarta, some 280km away.
But these claims were directly contradicted by Dr Mark Tingay from Curtin University, Australia, a geological pressure and rock mechanics expert.
The earthquake "was at least an order of magnitude too small," he said, stressing that the force felt at the Lusi site would have been "very small" - comparable to the effect of a heavy truck passing overhead.
When the vote was called, 42 out of the 74 scientists in the audience were convinced that the drilling was the trigger of the eruption.
Only three voted for the earthquake.
A further 16 scientists believed the evidence was inconclusive, and the remaining 13 felt that a combination of earthquake and drilling was to blame.
"The geologists voted overwhelmingly that drilling was the most likely cause," said Prof Underhill.
"The atmosphere was very tense, so all credit to them for not sitting on their hands.
"Hopefully this will be a catalyst for taking things forward. To my mind the result demonstrates that at the very least, the drilling company have a case to answer."
Prof Davies said: "I remain convinced that drilling was the cause of the mud volcano.
"The opinion of the international scientists adds further weight to my conviction."
Victims scatter flower petals on their former village, now swamped by mud
For two years, the Lusi crater has been oozing mud - enough to fill 50 Olympic size swimming pools every day.
The eruption began at 0500 on 29 May 2006 in the Porong subdistrict of Sidoarjo, Eastern Java, close to Indonesia's second city of Surabaya.
All efforts to stem the flow have failed - including a network of dams; channelling into the sea; and an ambitious plan to plug the crater with concrete balls.
Some geologists believe Lusi could continue to erupt for decades.
The mud flow has razed four villages and 25 factories. Thirteen people have died, as a result of a rupture in a natural gas pipeline underneath one of the holding dams.
Victim of Lusi mud volcano
The victims of the mud volcano
A police investigation is underway to identify the trigger and to determine whether the drillers are liable for compensating 10,000 families, amounting to 700 billion Indonesian Rupias (US$77; £47m).
If the earthquake is judged responsible, as claimed by Lapindo, then the Indonesian government will have the burden of supporting the victims.
There is no dispute that seismic activity can provoke mud volcanoes, and both are common in East Java.
Nevertheless, in June 2008 Prof Davies published a paper in the journal Earth and Planetary Science Letters, in which he concluded with "99% certainty" that Lapindo's drilling caused the mudflow.
He argues that the 2,500m-deep bore hole ruptured limestone rock, containing pressurised water. As the lower part of the borehole was not protected by casing, this forced water and mud into the rocks surrounding the well.
Homes swamped by mud
The volcano ejects enough mud each day to fill 50 Olympic swimming pools
At the conference, he produced fresh records of the changes in pressure in the 24 hours leading up to Lusi's eruption.
The pressure plots were introduced by drilling engineer Susila Lusiaga, who works with the Indonesian police investigation team.
"The pressure in the well went way beyond what it could tolerate... and it triggered the mud volcano," he said.
The new records "provide a compelling tape recording of the well as it started to leak," said Prof Davies.
"This is the data we wanted to get out - the data I have never been able to show before.
"It clearly shows that the well failed. And this failure was the driver for a the breakdown of the rocks - it was the trigger for the mud volcano."
Debate Chairman John Underhill is also a professional football referee
The well took a huge influx of fluid the day before the eruption, he said, resulting in intolerable pressures, and fractures which propagated until the surface was breached.
"We see the pressure building, then suddenly we see a massive drop at 9.30pm on May 28th - the night before the eruption began.
"This is evidence that a fracture has opened up. It's like a tyre bursting - the pressure inside bleeds away.
He added: "This may be evidence that Lusi actually started at 9.30pm the night before - not 5am the next morning."
Giant levees have been constructed to contain the mud
"Now the data has been released, I would like to get it out to independent drilling experts, who can then go through it," said Prof Davies, a geologist.
"We are particularly grateful to Lapindo, who were widely applauded at the meeting for their willingness to take part. We are now starting to make some headway."
However, despite the vote, the drilling firm strenuously denies that its activities were in any way responsible for the disaster.
From the same primary data, they calculate that the pressures under the ground did not go beyond critical levels.
"We presented clear and indisputable facts that none of the four required factors for the well to have been responsible for triggering the eruption occurred," a spokesman for Lapindo Brantas said.
"Specifically: there was no uncontrolled 'kick'. The casing shoe was not breached and the well was intact.
"There was no underground blowout. There was no sustained pressure to propagate a fracture."