Cement used to seal the Macondo well may have contributed to the blowout that caused the Gulf of Mexico oil spill, US investigators have found.
Both BP, which owned the well, and Halliburton, the contractor responsible for the cement, were aware of tests showing it was unstable, they said.
Halliburton has denied the claims, saying the tests were invalid as they were on a different kind of cement.
The 20 April blowout led to the worst environmental disaster in US history.
Eleven workers on the Deepwater Horizon drilling rig were killed by an explosion which followed, and hundreds of miles of coast were polluted.
The well was finally capped on 15 July, after an estimated 4.9m barrels of oil (171m gallons) had leaked into the sea, and fully sealed last month.
In a letter to the National Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling, its chief counsel Fred Bartlit said BP and Halliburton both had test results showing the foam slurry - created by injecting nitrogen into cement - used to seal the well before the blowout did not meet industry standards for stability.
Halliburton carried out four separate tests. In three of them, the results showed the mixture to be unstable and potentially prone to failure.
The fourth test produced more positive results, but these might not have been available by the time work on the well got under way on 19 April, Mr Bartlit wrote.
Finally, Halliburton and BP both had results in March showing that a very similar mix to the one pumped into the Macondo well would be unstable.
"But neither acted upon that data," Mr Bartlit wrote. "Halliburton (and perhaps BP) should have considered redesigning the foam slurry before pumping it at the Macondo well."
Halliburton later issued a statement saying it believed there were significant differences between its own tests and those performed by the commission.
"The commission tested off-the-shelf cement and additives, whereas Halliburton tested the unique blend of cement and additives that existed on the rig at the time Haliburton's tests were conducted," it said.
The company also denied any suggestion that it knowingly used a potentially defective product, suggesting that the failed internal tests were invalid because the cement mixture was different from the one eventually used in the well.
However, the last-minute changes demanded by BP meant that the final cement mix was not tested fully, it admitted. Earlier statements by the company had said tests had shown the cement to be stable.
Halliburton also criticised BP for not performing a cement bond log test, which checks the cement is secure after it has been pumped down the well, or properly interpreting the negative-pressure test.
"Given these numerous intervening causes, Halliburton does not believe that the foam cement design used on the Macondo well was the cause of the incident."
The BBC's Iain MacKenzie in Washington says the oil spill commission's findings go some way to supporting BP's own investigation, which found failings in the composition of the cement.
However, the full report has yet to be published and at this stage the investigators stop short of apportioning blame, our correspondent says.
Mr Bartlit also noted shortcomings by Transocean, which operated the Deepwater Horizon rig on behalf of BP.
"Because it may be anticipated that a particular cement job may be faulty, the oil industry has developed tests, such as the negative pressure test and cement evaluation logs, to identify cementing failures. It has also developed methods to remedy deficient cement jobs," he wrote.
"BP and/or Transocean personnel misinterpreted or chose not to conduct such tests at the Macondo well," he added.
But Mr Bartlit also cautioned in his letter that "the story of the blowout does not turn solely on the quality of the Macondo cement job".