In company filings published this week, it said 8.7% of its 955 million active accounts broke its rules.
Duplicate profiles – belonging to already registered users – made up 4.8% of its membership figure.
User-misclassified accounts amounted to 2.4% – including personal profiles for businesses or pets – while 1.5% of users were described as “undesirable”.
The estimate comes at a time of growing concern about the effectiveness of marketing on the platform.
Facebook defined duplicates as “an account that a user maintains in addition to his or her principal account.”
It said profiles were “user-misclassified” if “users have created personal profiles for a business, organisation, or non-human entity such as a pet”.
It added that “undesirable” accounts included those using fake names which were “intended to be used for purposes that violate our terms of service, such as spamming”.
‘Harm our business’
Facebook, whose business model relies on targeted advertising, is coming under increased scrutiny over the worth of its advertising model which promotes the gathering of “likes” from users.
“We generate a substantial majority of our revenue from advertising,” the company said in its filing.
“The loss of advertisers, or reduction in spending by advertisers with Facebook, could seriously harm our business.”
Last month, the BBC’s technology correspondent Rory Cellan-Jones set up a fake company called VirtualBagel to investigate allegations of fake “likes”.
His investigation found that the large majority of “likes” for the fake firm originated from the Middle East and Asia.
Many users appeared to be false, such as “Ahmed Ronaldo” – apparently a Cairo-based user who is employed by Spanish football club Real Madrid.
Last week, digital distribution firm Limited Press alleged that, based on its own analytics software, 80% of clicks on its advertisements within Facebook had come from fake users.
In a post on its Facebook page, the company said: “Bots were loading pages and driving up our advertising costs. So we tried contacting Facebook about this. Unfortunately, they wouldn’t reply.
“Do we know who the bots belong too [sic]? No. Are we accusing Facebook of using bots to drive up advertising revenue. No. Is it strange? Yes.”
After a surge of attention to the company, it has since removed the Facebook posting, and said Facebook was now looking into its concerns.