Twitter Inc said that its internal controls were allowing it to weed out accounts being used for the "promotion of terrorism" earlier rather than responding to government requests to close them down.
U.S. and European governments have been pressuring social media companies includingTwitter, Facebook Inc and Alphabet Inc's Google to fight harder against online radicalisation, particularly by violent Islamist groups.
Twitter said it had removed 299,649 accounts in the first half of this year for the "promotion of terrorism", a 20 percent decline from the previous six months, although it gave no reason for the drop. Three-quarters of those accounts were suspended before posting their first tweet.
Britain's interior minister, Amber Rudd, used a visit to Silicon Valley last month to ask Facebook, Microsoft, Twitter, and YouTube to step up efforts to remove content that incites militants after four attacks in Britain killed 36 people this year.
"Loser terrorists must be dealt with in a much tougher manner. The internet is their main recruitment tool which we must cut off & use better!" U.S. President Donald Trump tweeted on Friday after a bombing on a London commuter train.
Less than 1 percent of account suspensions were due to government requests, Twitter said, while 95 percent were thanks to the company's internal efforts to combat radical content with "proprietary tools", up from 74 percent in its last twice-yearly transparency report.
Twitter defines "promotion of terrorism" as actively inciting or promoting violence "associated with internationally recognised terrorist organisations."
The vast majority of notices from governments concerned "abusive behaviour", which includes violent threats, harassment, hateful conduct and impersonation.
Twitter said it had removed 935,897 accounts for promotion of terrorism between August 1, 2015 and June 30 this year.
The social media platform said in July it had 328 million average monthly active users in the three months to June 30.
The European Union has threatened legislation on Internet firms removing illegal content if they do not step up efforts to police what is available on the web.
Twitter said it had received about 3 percent more legal requests and court orders to remove content posted by users in the first half of this year than during the last six months of 2016.
About 90 percent of those removal requests came from Turkey, Russia, France and Germany.
The transparency report showed Turkey was the most active country in seeking the removal of content, accounting for 45 percent of all requests worldwide.
Twitter said it had received eight requests from governments to take down content posted by journalists and news organisations in the first half of 2017 but did not act on any of them "because of their political and journalistic nature."
Of the eight, five were court orders or other legal demands from Turkey ordering Twitter to take down content from journalists or news outlets.
Turkey detained tens of thousands of people including scores of journalists after a failed coup in July last year. The crackdown by Turkish President Tayyip Erdogan, who has for years tried to stamp out what he sees as illegal online activity, has strained relations with NATO allies and raised alarms among civil liberties advocates.
Twitter said it filed legal objections to court orders involving Turkish journalists and news outlets wherever possible but none of them had prevailed.
