Direct answer
A 22% Turnitin score is usually not bad, but the verdict depends entirely on which number you are reading. If 22% is your similarity index, it sits in the normal range for an essay that quotes sources and lists references, and most institutions do not treat it as misconduct. If 22% is your AI writing score, it is above Turnitin's 20% confidence threshold, which means the detector is reporting a real flagged portion rather than a low-confidence asterisk — and that is worth investigating before you submit [1]. The number alone never decides your case; what the matched or flagged text actually is decides it.
What Does A 22% Turnitin Score Actually Mean?
A Turnitin similarity index is the share of your submission that matched text in the comparison repository — it is a measure of overlap, not a measure of wrongdoing [2]. Turnitin states plainly that the percentage is not a plagiarism score and cannot be interpreted in isolation, because a paper with a well-built reference list and long block quotations can reach the low twenties without a single uncited sentence. The report exists to point a reader at passages worth looking at, and the match overview is where that inspection happens.
Where the 22% comes from matters more than the figure itself. If most of the overlap is concentrated in your bibliography, in direct quotations inside quotation marks, or in a handful of short generic phrases such as "the results of this study suggest", the number is inflated by material your marker already expects to see [2]. If instead a single source accounts for 14 of those 22 points across two uninterrupted paragraphs of body text, you are looking at a genuine attribution problem rather than a formatting artefact.
The AI side of the report works on a different scale entirely. Turnitin's AI writing indicator estimates the proportion of qualifying prose that its model attributes to generative tools, and it is reported separately from similarity [1]. Turnitin also suppresses the exact figure and shows *% when its confidence falls below 20%, so a visible 22% AI score is a signal the system considers confident enough to display — not a borderline artefact [1]. Reading the two numbers as if they were the same metric is the single most common misreading of a Turnitin report.
One practical consequence follows: never quote "22%" to a tutor without saying which report it came from. A similarity figure and an AI figure carry different evidence, different defences, and different remedies, and conflating them weakens the strongest thing you have — an accurate account of your own drafting process [1].
Is 22% Similarity Index Bad Enough To Be Penalised?
There is no universal cut-off, because acceptable similarity is set by your institution and often by your department or individual module handbook, not by Turnitin [3]. Some departments flag anything above 20% for review; others routinely accept 30% on literature reviews where quoting is unavoidable. The figure that matters is the one written into your assessment guidance, and if that guidance is silent, the honest position is that you do not know your threshold yet — so ask before you panic or before you resubmit.
The causes of an elevated score are usually mundane, and Turnitin's own support guidance lists them: a reference list that matches every other student's reference list, correctly cited quotations, appendices, and short phrases that appear in thousands of papers [3]. Running the report with quotes and bibliography excluded frequently drops a 22% down to single digits, which tells you the overlap was never substantive. That exclusion view is a diagnostic tool, not a defence — you cannot submit an excluded report as your result — but it tells you quickly whether you have a real problem.
Where 22% becomes serious is when the matches sit in your analytical body text and are not attributed. Two or three sentences lifted from a source without quotation marks and a citation is a citation error at best and an integrity matter at worst, and the percentage is irrelevant to how that is handled [3]. Markers read the highlighted passages, not the headline number, so a 22% made of properly attributed material is a far safer position than a 9% containing one unattributed paragraph.
If you have already received a report and are unsure how it will be read, the productive move is to open each match and classify it yourself: quotation, reference, common phrase, or unattributed borrowing. That classification is exactly what an academic integrity officer will do, and doing it first puts you in a position to explain your draft rather than defend it [3].
How Can You Check Whether Your 22% Will Be Flagged Before You Submit?
The most reliable way to know how your 22% will land is to see the same report your instructor will see, on your own draft, before the deadline. Turnitin's own guidance to institutions stresses that AI and similarity indicators are meant to be read alongside context about how a piece was drafted, not as a standalone verdict [4]. That cuts both ways: it means a 22% is rarely fatal on its own, and it also means you cannot assume a marker will give you the benefit of the doubt without evidence.
Evidence is the part students most often skip. If you drafted with ChatGPT, Claude, or Gemini and then edited by hand, keep your version history, your prompt log, and your notes — because a 22% AI score on edited AI text is a scenario where documentation decides the outcome [4]. Turnitin acknowledges that detection can produce false positives, particularly on highly formulaic academic prose and on writing from non-native English speakers, and institutions are expected to have a process for discussing a flagged report with the student rather than acting on the number alone [4].
A pre-submission check closes the gap between guessing and knowing. Running your draft through a Turnitin check returns the AI detection report and the similarity report as two PDFs, showing the flagged passages and matched sources exactly as they appear in the institutional view. At 22% you would immediately see whether the flags cluster in one section you can rewrite, or whether they are spread thinly across the whole document — two very different problems with two very different fixes.
Timing is the other variable worth controlling. Discovering a 22% AI score the night before submission leaves you no room to rewrite flagged paragraphs properly, whereas finding it a week out lets you revise the specific passages, re-check, and confirm the number has moved. The check itself is fast — most orders complete in 5 to 15 minutes — so the constraint is your editing time, not the report turnaround.
If your 22% came from a detector you do not fully trust, or from a free tool that never showed you the flagged passages, the sensible next step is to look at the real report before you make any decisions about your draft. turnitin0 exists for exactly that moment: it returns the actual Turnitin AI detection report and similarity report on your own document, so you can see which passages produced your 22% and decide whether to revise, cite more carefully, or submit with confidence.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does a 22% Turnitin score mean I plagiarised?
No. The similarity index measures matched text, not misconduct, and a 22% is commonly produced by reference lists, quotations, and short common phrases [2]. Whether anything is wrong depends on whether the matched passages are properly attributed.
Is 22% AI score high on Turnitin?
It is above Turnitin's 20% confidence threshold, so the figure is displayed rather than replaced by *% — meaning the detector considers the signal meaningful [1]. Whether it is "high" depends on how much of your document the flagged passages cover.
Can a 22% similarity score get me a zero?
Only if the underlying matches show unattributed borrowing, because institutions penalise the passages rather than the percentage [3]. A 22% built from cited material is not grounds for a penalty at most institutions.
Why did my score jump from 12% to 22% after resubmission?
Resubmitting can add your paper to the comparison database, so later drafts may match earlier versions of your own work [3]. Check whether the new matches are against your own previous submission before assuming anything else.
Should I rewrite my whole essay at 22%?
Usually not. Open the report, identify which passages actually produced the matches or flags, and revise only those — rewriting everything risks introducing new problems and wastes time you could spend on the flagged sections [4].