The copy of a book that trips Turnitin is usually not the book. It is a classmate's paper from three years ago that quoted the same paragraph.
That is the part most students have backwards. Turnitin does not hold a full-text library of everything ever published, and its own marketing does not claim one. What it holds is web pages, journal content from publishing partners, and a very large pile of papers other students have already submitted, and the third of those is the one doing most of the work on any book you are worried about.
Here is what is actually in there, source by source, and what is genuinely missing.
The three things it searches
Turnitin's own guides describe a similarity report as a comparison against the search targets selected for that assignment, and it names three: current and archived internet content, previously submitted student work, and tens of thousands of periodicals, journals, and publications.
On the scale of the first one, its database-content page says the crawler has indexed 54 billion web pages, and that it crawls current and archived pages. The archived part is the one worth pausing on. A page that was deleted two years ago can still produce a match, because the copy Turnitin keeps is its own.
For the publication tier, the same page names its partners: Elsevier, Springer Nature, Wiley, Taylor and Francis, IEEE, and open access journals. That is a journals list, not a library catalogue, and the distinction is exactly the point of the next section.
One detail that gets missed: those search targets are settings. An instructor can enable or disable them per assignment, so two assignments at the same school are not necessarily checked against the same database.
Books: mostly no, and the exception is the whole story
Turnitin does not advertise a full-text book corpus, and university writing centres that support the tool are blunter about it. The University of Maryland Global Campus tells its students that the Turnitin database does not search books or articles in subscription databases available through its library, and that matching text from those sources will not surface unless the material also appeared in assignments previously sent to Turnitin.
Read that exception again, because it is the mechanism that matters. A passage from a well-known book does not need to be in the database as a book. It needs one student, once, to have quoted or patchwritten it in a submitted paper. After that the text is in the repository, and every future submission is compared against it.
This produces a result that feels backwards but is consistent: the more heavily a book is assigned, the more likely a sloppy paraphrase of it gets flagged, because popularity on syllabi is what puts it in the index. An obscure monograph nobody in your program has cited is far less exposed than the chapter everyone in your seminar was assigned last spring.
Lecture slides and course handouts
These come down to one question: can a crawler reach it?
Material sitting behind your learning management system's login is not in any web index. Google's own documentation is explicit that password-protecting a page prevents Googlebot and other web crawlers from accessing it, and Turnitin's crawler is subject to the same wall. A slide deck posted only inside your LMS is not something the web tier can see.
Change one variable and the answer flips. Slides on a public department page, a conference deck uploaded to a slide-sharing site, or a PDF a student mirrored somewhere are all ordinary crawlable pages.
And the repository route works here too. If a classmate lifted three sentences from the same slide last year and submitted it, those sentences are in the student paper index regardless of where the original deck lives.
Old student papers, including your own
The student repository is the tier people underestimate most, and it has a setting attached that students almost never think to ask about.
Turnitin's assignment options let the instructor choose where a submission is stored. The standard paper repository is the global one, which means the paper is compared against submissions from other institutions. An institution repository keeps papers private to your school, so they are only compared with other papers from there. And the no-repository option generates a similarity report without storing the paper for future comparison at all.
That third setting exists for a specific reason, and it is the one to care about. If you submit a draft to an assignment configured for the standard repository, that draft is now in the index. Your final version, which is mostly the same text, gets compared against it and matches heavily. The result is a terrifying score produced entirely by your own earlier work.
So before you submit a draft anywhere, ask which repository setting the assignment uses. Instructors who set up practice submissions properly use no repository for exactly this reason, but you cannot tell from the student view.
Homework-help and study sites
Content on sites like Chegg and Course Hero reaches the database through the ordinary web tier when the pages are publicly reachable, and does not when they sit behind a paywall or an account. Which is why the answer to whether an old solution will match is genuinely "it depends on the page", and why identification usually comes from somewhere other than the similarity report.
What is definitely not in there
Being specific about the gaps is more useful than a vague reassurance that the database is enormous.
- Articles in paywalled subscription databases your library licenses, unless a previous submission carried the text in
- Anything behind a login, including course materials, private forums, and gated archives
- Print-only material that was never digitised
- Papers submitted to assignments configured with no repository
- Papers held in another institution's private repository
That list cuts in both directions, and the second direction is the one worth internalising. A clean similarity report does not prove a paper is original. It proves the paper did not match the specific corpus that was searched, which is a much smaller claim.
Why another checker gives you a different answer
Once you see the database as a set of choices rather than a single universal index, disagreement between tools stops being mysterious. The corpus is the product. Two checkers with different corpora, different crawl histories, and different filters will read the same paper differently, which is most of what is going on when the same paper scores differently in SafeAssign and Turnitin.
The same honesty applies to us. PlagiarismCheck.io searches web content and gives you source links for every match, and it has no access to the student paper repository, because nobody outside Turnitin does. That makes it useful for catching the web-facing matches before you submit, and useless for predicting whether your paper will hit a classmate's essay from 2023. Any tool claiming to check your work against Turnitin's student repository is not doing that.
What this changes about how you write
Three practical consequences follow from all of this.
Popular sources deserve more care, not less. The chapter everyone in your cohort was assigned is the single most indexed thing you will cite this term, and a lazy paraphrase of it is the likeliest match in your paper.
Deleting a source does not help anybody, since the archived crawl outlives the live page. And your own drafts are a real risk that has nothing to do with dishonesty, which is worth one email to your instructor before you upload one.
The short version
Does Turnitin check books? Not as books, in most cases. It checks the web, its publishing partners' journals, and every paper it has ever been given, and books leak into that third pile one student quotation at a time.
Which means the useful question is not what is in the database. It is whether the specific sentences you wrote could have passed through someone else's paper first. If you want to see the web-facing half of that picture before you submit, PlagiarismCheck.io is free, needs no account, and shows you the source behind every match it finds.