Direct answer
The Turnitin similarity index is the percentage of a submission's text that matches sources in the Turnitin repository, and it is not a plagiarism score [1]. Turnitin itself publishes no universal "safe" threshold: what counts as acceptable is set by your institution and your assignment, so the number only means something when you read the match list behind it [2]. Because the repository keeps growing and instructor settings change, the same document can return a different index on a later run [3]. Students cannot run a general Turnitin check on demand — institutional draft assignments are the only official route, which is why independent pre-submission checkers exist [4].
What Does the Turnitin Similarity Index Actually Measure, And What Counts As A "Good" Percentage?
The similarity index is a simple ratio expressed as a percentage: the number of words in your document that Turnitin matched against its comparison sources, divided by the total word count of the document [1]. Those sources fall into three buckets — internet pages, publications, and previously submitted student papers — and the report breaks the index down by bucket so you can see where the overlap originates [1]. Crucially, the index aggregates all matched text, which means correctly formatted quotations, paraphrased material with attribution, and even a long reference list all push the number upward [1]. A high index therefore tells you that text is duplicated somewhere, not that you have done anything wrong.
Turnitin's own guidance is explicit that no fixed cut-off separates acceptable from unacceptable work [2]. Expectations are set locally: a law school, a nursing programme, and a first-year history module can all treat the same percentage very differently, and a lab report full of standard methodology phrasing will naturally score higher than an original critical essay [2]. The company advises markers to look at what matched rather than the headline figure, because a 28% index built from a reference list and four short quoted phrases is a completely different document from a 28% index built from three unquoted paragraphs [2].
The practical way to read your own report is to work through the match list in order of size. Open the largest matches first and ask a single question of each: is this text properly attributed and correctly formatted? If it is, the match is expected and defensible; if it is not, that is the passage to rewrite in your own words [1]. Reviewers reading the same report will follow exactly this path, so rehearsing it yourself is the most reliable preparation you can do [2].
One detail worth knowing: Turnitin reports a small amount of matching on almost every academic document, because titles, module codes, standard terminology, and bibliographic entries are shared across thousands of papers [1]. Treating a low single-digit or low double-digit index as a red flag is a misreading of the metric. The signal that matters is unattributed overlap with a source you did not cite [2].
Why Does The Same Paper Get A Different Similarity Index Each Time It Is Checked?
The comparison repository is not a static database — it grows continuously as new student papers are ingested, new publications are licensed, and new web pages are crawled [3]. A document checked in week one can match a paper submitted in week three, so a re-run of an unchanged file can legitimately return a higher index [3]. This is one of the least understood properties of the system, and it explains most of the "my score went up and I changed nothing" messages that circulate before deadlines [3].
Instructor settings are the second variable. Markers can switch exclusions on or off for quotations, the bibliography, and matches below a word-count threshold, and each combination changes the reported index without a single character of your text changing [3]. A report generated under one set of exclusions is not directly comparable to a report generated under another, even for the identical file [3].
The third variable is the repository itself. A check run against a different corpus, or with different settings, will not reproduce an institutional number exactly, and that applies to any third-party service as much as to a departmental draft folder [3]. If you are comparing two reports, compare the match lists and the settings behind them rather than the two headline percentages [3].
The sensible conclusion is to treat the index as a diagnostic snapshot taken at a specific moment under specific settings, not as a fixed property of your document [3]. What you want from a pre-submission check is the match list — the passages, the sources, and the word counts — because that is the part that stays meaningful even when the percentage shifts [3].
Can I See My Own Turnitin Similarity Index Before I Submit?
Turnitin does not offer a general student-facing checker that anyone can use on demand; student access depends entirely on whether the institution has enabled a draft or self-check assignment in the course [4]. Where a lecturer does enable one, the draft is typically excluded from the student paper repository, so checking your own draft does not create a future match against your own work [4]. Where no such assignment exists, there is no official route for a student to generate a similarity report independently [4].
That gap is why independent pre-submission services exist, and it is worth being precise about what they can and cannot do. A service that runs a genuine Turnitin check returns the same report format your marker sees — similarity summary, source breakdown, and match highlights — but it is not affiliated with Turnitin, and its output should be read with the repository caveats above in mind [4]. What such a check reliably gives you is the match list: the specific passages and sources that are driving your index, which is exactly the information you need in order to edit [4].
Used well, a pre-submission check is a revision tool rather than a verdict machine. Run it once, open every match above a couple of percent, fix the passages that are unattributed, then re-run to confirm the index has moved in the direction you expect [4]. Students who do this usually discover that the bulk of their index comes from two or three fixable passages — a definition lifted from a textbook, a methods paragraph copied from a lab manual, a quotation missing its citation [4].
A word of caution about timing: because the repository and the settings both move, a check run days before a deadline is more informative than one run weeks earlier [3]. And because a draft check may be excluded from the student database while a final submission is not, the number you see on a draft is a close estimate of the institutional result, not a guarantee of it [4].
If you would rather see the actual report than guess at the number, turnitin0 gives you the same similarity summary and match highlights your marker will open — with the AI writing report delivered alongside it, in under fifteen minutes for most orders and without adding your file to the student paper database.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a 20% similarity index bad in Turnitin?
Not on its own. Twenty percent of matched text can come entirely from a reference list and correctly cited quotations, or it can come from unattributed paragraphs — only the match list distinguishes the two [1]. Turnitin publishes no universal threshold, so the relevant standard is your institution's [2].
Why is my similarity index different from my friend's for the same assignment?
The index reflects each document's own word count and each report's exclusion settings, so two papers on the same topic rarely produce identical numbers [3]. Repository growth between the two check times can also shift the result [3].
Does Turnitin's similarity index detect AI writing?
No. The similarity index measures text matching against sources, while AI writing detection is a separate report with its own score and threshold [1]. A low similarity index says nothing about whether a document will be flagged for AI-generated content.
Can I check my similarity index without adding my paper to the student database?
Institutional draft assignments are usually excluded from the student paper repository, so a draft check does not create a match against your own later submission [4]. Independent pre-submission checks operate outside Turnitin's repository and should be treated as an estimate of the institutional result [4].
What should I do if my similarity index is higher than I expected?
Open the largest matches and check attribution first, then rewrite any unattributed passage in your own words and re-run the check [1]. Because the repository shifts over time, run the final check close to your deadline rather than weeks in advance [3].