PRISMA Flow Diagram Generator
Paste your screening counts — or the whole methods section — and get a PRISMA 2020 flow diagram with the official box labels, your exact numbers, and the arithmetic checked before anything is drawn.
One review, four styles
The same screening counts rendered in each style — click a card to use that look.
What is a PRISMA flow diagram?
A PRISMA flow diagram is the figure that documents how a systematic review went from everything the search returned to the studies that were finally included. The PRISMA 2020 statement defines it exactly: three phases — Identification, Screening and Included — with standardized boxes recording how many records were identified, how many were removed before screening, how many were screened and excluded, how many reports were sought, not retrieved and assessed, and how many studies made it into the review. Checklist item 16a asks authors to report this selection process "ideally using a flow diagram", and most journals treat that as a requirement for systematic reviews and meta-analyses.
This PRISMA flow diagram generator does the assembly for you. Describe your screening in plain words, paste the numbers in any order, or attach the review itself — the AI maps your counts onto the official PRISMA 2020 template, keeps every box label verbatim, fills in only what the arithmetic determines, and flags any numbers that do not reconcile instead of silently changing them. Then it renders the diagram in the style you pick: the austere official look, a softly tinted version, a bold slide-deck variant, or a whiteboard sketch for teaching.
Key facts
- PRISMA 2020 replaced PRISMA 2009; the current diagram has three phases — Identification, Screening and Included — not the four of the old version.
- There are two official templates: one for reviews of new searches only, and one that also reports studies carried over from a previous version of the review.
- Records, reports and studies are three different units, and mixing them is the most common reason a diagram fails to reconcile.
- Every exclusion at full-text stage must carry a reason and a count; a bare total is not compliant.
- The box labels are fixed by the statement — a diagram that renames them is no longer a PRISMA diagram.
How does the PRISMA flow diagram generator work?
Paste your counts
A methods paragraph, a messy list of numbers, or the full PDF — records, reports, exclusion reasons, in any order. No form with fifteen fields to fill.
AI builds the box list
Your counts are mapped onto the official PRISMA 2020 boxes, labels verbatim. The three reconciliation checks run first — mismatches are flagged, never silently corrected.
Render and refine
Get the diagram in the official look or a styled variant, then edit any text in the canvas or upscale it for submission.
Every box in the PRISMA 2020 flow diagram
These are the boxes of the official PRISMA 2020 template for a new review searching databases and registers, with their labels exactly as the template prints them. The generator uses these same labels — and the tool page below each box explains which number belongs in it:
Identification
One count per source type, as returned by the searches: database records and register records, before any deduplication.
Everything discarded before a human screened titles and abstracts — duplicates, records an automation tool ruled ineligible, and any other pre-screening removals.
Screening
Records whose titles and abstracts were actually screened. Equals records identified minus records removed before screening.
Records rejected at title-and-abstract stage. PRISMA does not require reasons here — one total is enough.
Records that survived screening, now pursued as full-text documents. From here the diagram counts reports, not records.
Full texts you tried to obtain but could not — paywalls, dead links, no response from authors.
Full-text reports read against the eligibility criteria. Equals reports sought minus reports not retrieved.
Reports rejected at full-text stage — here PRISMA does require a reason and a count for each, e.g. wrong population (n = 12), wrong study design (n = 8).
Included
The end of the flow: how many studies are in the review, and how many reports describe them. One study can span several reports, which is why both counts appear.
Reviews that also searched websites, organisations or reference lists add a second identification column — "Identification of studies via other methods" — with its own retrieval and eligibility boxes. Updated reviews add a column for studies carried over from the previous version. Delete boxes for stages you did not use; do not fill them with n = 0.
The three arithmetic checks
A methods reviewer reconciles the diagram before reading anything else. These are the three equations the numbers must satisfy — the generator runs them on your counts and flags any mismatch:
Identification balances
Records identified − Records removed before screening = Records screened
Everything identified either left before screening or was screened. If this fails, a deduplication count is usually missing or double-counted.
Screening balances
Records screened − Records excluded = Reports sought for retrieval
Every screened record was either excluded on title and abstract or pursued in full text. A gap here usually means an exclusion total was left out.
Retrieval balances
Reports sought − Reports not retrieved = Reports assessed for eligibility
Every report you sought was either retrieved and assessed or could not be obtained. From this point the diagram counts reports — one study can have several.
Where are PRISMA flow diagrams used?
Systematic reviews & meta-analyses
The figure journals expect with checklist item 16a — official labels, your counts, portrait layout ready for the methods section of the paper or the supplement.
Scoping reviews & theses
PRISMA-ScR and student reviews use the same 2020 flow. Paste the numbers from your screening spreadsheet and the diagram is ready for the committee.
Teaching & protocols
The whiteboard and presentation styles explain the selection process in lectures, protocol meetings and journal clubs — same rigor, friendlier look.
PRISMA 2020 vs PRISMA 2009
PRISMA 2020 is the current standard and most journals now require it. The visible differences are structural: the 2009 diagram had four phases and a single "records screened" path, while 2020 has three phases, separates database from non-database sources, and reports removals before screening explicitly. A review submitted today with a 2009 diagram will usually be asked to redraw it.
| PRISMA 2009 | PRISMA 2020 | |
|---|---|---|
| Phases | Identification, Screening, Eligibility, Included | Identification, Screening, Included |
| Search sources | One combined path | Databases and registers split from other methods |
| Duplicate removal | Implied in "records screened" | Reported explicitly before screening |
| Exclusion reasons | At full-text stage | At full-text stage, with counts per reason |
| Previous review versions | Not covered | Second template covers updates |
| Term used for the unit | "Articles" used loosely | Records, reports and studies defined separately |
If your protocol was registered against PRISMA 2009 you may still report against it, but say so in the methods — silently mixing the two structures is what reviewers actually object to.
Common questions about PRISMA flow diagrams
What is a PRISMA 2020 flow diagram?
It is the standardized figure defined by the PRISMA 2020 statement that documents study selection in a systematic review: how many records the searches identified, how many were removed before screening, how many were screened, excluded, sought as full texts, assessed for eligibility, and how many studies were finally included. The boxes, their labels and their order come from the official template.
Is a PRISMA flow diagram required for a systematic review?
PRISMA 2020 checklist item 16a asks authors to describe the search and selection results "ideally using a flow diagram", and most journals that require PRISMA compliance treat the diagram as mandatory in practice. Reviewers routinely check that the counts reconcile across the boxes.
Which PRISMA template does the generator use?
The PRISMA 2020 template for new reviews searching databases and registers, which covers the large majority of reviews. If your data mentions websites, organisations or citation searching, the second identification column — "via other methods" — is added automatically. Scoping reviews (PRISMA-ScR) use this same 2020 flow.
What is the difference between records, reports and studies?
A record is one search hit — a title and abstract from a database. A report is the retrieved document itself. A study is the piece of research, which may be described in several reports. The flow counts records through screening, switches to reports at retrieval, and ends with both studies included and reports of those studies.
What if my numbers do not add up?
The generator runs the three reconciliation checks — identified minus removed equals screened, screened minus excluded equals sought, sought minus not retrieved equals assessed — before drawing. A count the arithmetic fully determines is filled in; a genuine mismatch in your numbers is kept verbatim and flagged in the analysis so you can fix the source, not the figure.
Can I edit the diagram after it is generated?
Yes. Every generation opens in the workspace, where Edit in Canvas lets you change any label or count, and the upscaler exports a high-resolution version for submission. Verify the counts against your screening log before you submit — the diagram is a rendering of your numbers, and the numbers are yours.
How do you fill in a PRISMA flow diagram?
Work top to bottom and record every number you actually observed, not the ones that make the arithmetic tidy. Start with records identified per database and per register, then removals before screening, then records screened and excluded, then reports sought and not retrieved, then reports assessed for eligibility with exclusions itemised by reason, and finally studies included. Each stage carries its own count, and the differences between stages must be accounted for.
What is the difference between PRISMA 2020 and PRISMA 2009?
PRISMA 2020 has three phases where 2009 had four, splits database and register searches from other search methods, and reports pre-screening removals explicitly. It also defines records, reports and studies as distinct units. PRISMA 2020 is the current standard and is what most journals now ask for.
Do I need to give a reason for every excluded full text?
Yes. At the eligibility stage the statement requires exclusions to be itemised — each reason with its own count, adding to the total excluded. "Excluded (n = 42)" without reasons is the single most common compliance failure. Reasons are usually short: wrong population, wrong intervention, wrong outcome, wrong study design, no full text available.
Which template should I use if my review is an update?
Use the second official template, the one with columns for studies carried over from the previous version. It reports new records alongside those already included in the earlier review, so a reader can see what the update actually added. Using the new-searches-only template for an update hides that distinction.
Where do records found by hand searching or citation chasing go?
In the right-hand column, under searches via other methods — separate from databases and registers. That separation is one of the main changes PRISMA 2020 introduced, precisely so readers can see how much of the yield came from formal database searching and how much from other routes such as citation chasing, contacting authors, or grey-literature sources.
Is a PRISMA flow diagram needed for a scoping review?
Scoping reviews use PRISMA-ScR, an extension with its own checklist. The flow diagram itself is essentially the same shape, so the diagram you build here works for a scoping review too — but the accompanying checklist you report against should be PRISMA-ScR, not the main PRISMA 2020 checklist.
Can the numbers be corrected after the diagram is drawn?
Yes — the diagram opens in a canvas workspace where any count can be edited directly, and the three arithmetic checks re-run against the corrected figures. That matters in practice because screening counts usually change once or twice during peer review, and redrawing the whole figure by hand each time is where transcription errors get introduced.
Sources
- The PRISMA 2020 statement: an updated guideline for reporting systematic reviews — BMJ 2021;372:n71
- PRISMA 2020 flow diagram — PRISMA statement
Keep exploring
Our other diagram tools, and the guides behind them.
Tools
Your counts are ready. The diagram takes a minute.
Paste the numbers from your screening log and get a reconciled PRISMA 2020 flow diagram in the style your venue needs.
Generate your PRISMA diagram