The US government wants to make sure banks don't get into trouble again
By Greg Wood
North America business correspondent, BBC News
The stress tests conducted on the 19 largest US banks have found that 10 of them require a combined $74.6bn (£50bn) of additional funds.
But how reliable have the tests been and what have they achieved?
"Stress tests are more of an art than a science," says Peter Vinella, a former mathematics professor and now head of financial services at LECG, a consulting firm which helped one of the big US banks through the testing process.
Banks have been conducting stress tests for years, to see how their assets would be affected by a particular event.
For example, how many of their customers would default on their home loans if interest rates went up by 1%?
But the tests are based on assumptions, or "variables", as mathematicians call them. And that means there's always a margin of error.
"A lot of the time when people think of a mathematical formula, they think in terms of the laws of physics," says Mr Vinella.
"Nothing could be further from the truth.
"Even though there are mathematical models, these are all assumptions that are very difficult to prove one way or the other."
The stress tests conducted on the banks made a whole range of assumptions, about future economic growth, the level of unemployment and where house prices are going. Mr Vinella says that makes the results even more unreliable.
"The more factors you start adding into a formula, the less explanatory power each of those factors has. That's because they're not independent of each other.
"As you add in more and more factors, you get more overlap and the accuracy of the model actually goes down."
Pass or fail?
One of the assumptions made in the stress tests is that the US economy will perform even worse than expected at the moment, with unemployment rising to more than 10% of the workforce by the end of next year.
On the basis of what may happen in that worst case scenario, some banks are being asked to raise new capital now. The banks think that is unfair.
"It's ill-advised to say to a bank that you're going to hold them to a more adverse set of circumstances two years out," says Mark Tenhunfeld, from the American Bankers' Association, "and judge whether or not they have sufficient capital today.
"The regulators, in their efforts to do good, may cause banks to set aside capital that will not be needed."
Whether the authorities like it or not, US banks will now be divided into those who have "passed" the stress tests and those who have "failed" and need to raise new capital. The question is - where will they get it?
Private investors may well be reluctant to pump new money into "failed" banks, forcing them to go to the government for more financial support. That's unlikely to boost confidence in the long-term future of those banks.
When the stress tests were announced back in February, there were real concerns that some banks would have to be nationalised.
That's been averted, and since then the financial sector has shown signs of recovery. The stress tests have been successful as a PR exercise. Whether they will actually make the weak banks any stronger is debatable.