US President-elect Donald Trump has agreed to settle fraud lawsuits relating to his Trump University series of real estate seminars for $25 million, New York's attorney general has said.
A settlement ends a dispute that dogged Mr Trump throughout his presidential election campaign and led to one of the more controversial moments of his run when he claimed the judge overseeing two of the cases was biased because he was of Mexican ancestry.
Lawyers for the president-elect have been arguing against students who claim they were they were lured by false promises into paying up to $35,000 to learn Mr Trump's real estate investing "secrets" from his "hand-picked" instructors.
There were three lawsuits relating to Trump University: two class actions suits in California and a case brought by New York Attorney General Eric Schneiderman.
All of the cases were covered in the settlement.
"Today's $25 million settlement agreement is a stunning reversal by Donald Trump and a major victory for the over 6,000 victims of his fraudulent university," Mr Schneiderman said in a statement.
Lawyers for Mr Trump, who has denied any wrongdoing, could not immediately be reached for comment.
However Mr Trump responded on Twitter: "I settled the Trump University lawsuit for a small fraction of the potential award because as President I have to focus on our country."
Class action lawyers for the students were not planning to seek attorneys fees, but reimbursement for costs, a source familiar with the situation said.
US District Judge Gonzalo Curiel, who was presiding over the two California cases, had urged both sides to settle.
Mr Trump said during his election campaign that Judge Curiel, who was born in Indiana to Mexican immigrant parents, could not be impartial because of Mr Trump's campaign pledge to build a wall on the US-Mexico border to control illegal immigration.
Mr Trump has said he did not "hand pick" Trump University instructors, but that marketing language used was not to be taken literally.
He has said most students gave the classes high ratings.
