PROSPERO Explained: Why Protocol Registration Matters
PROSPERO is a register, not a gatekeeper: its checks “do not constitute peer review or imply approval of the methods proposed”, and CRD’s guidance sets two separate timing thresholds that the widely repeated advice to register before data extraction begins matches neither of.
- Which review types PROSPERO accepts, and the two it turns away outright
- The two different timing thresholds in CRD’s guidance, and which one is the eligibility limit
- What registration does and does not certify about your methods
- “Register before data extraction begins.” This is repeated almost everywhere, and it matches neither threshold CRD actually publishes. CRD’s recommended point is earlier — before screening against eligibility criteria commences — and the eligibility limit it states is later, at the completion of data extraction. Treating the halfway point as the rule means some authors register later than best practice, while others abandon a registration they were still entitled to make.
What PROSPERO is
PROSPERO is a free, open register of systematic review protocols run by the Centre for Reviews and Dissemination (CRD) at the University of York and funded by the UK’s National Institute for Health and Care Research. It has operated since 2011 as a platform for “registrations of systematic review protocols with a direct health-related outcome”.3
Its scale is now the main thing that distinguishes it from other registries. CRD reported 306,444 published records as of 11 December 2024, against 359 records in PROSPERO’s entire first year. In November 2024 alone, 6,458 records were published — roughly 215 a day — and PROSPERO gained more than 7,000 new registered users in that single month, around seven times its first-year user total.3
That growth matters practically. A register this dense is the first place to find out whether the review you are planning is already under way somewhere else, which is the use CRD itself now designs for.
Registration starts from “Register a review”. The home page states who can register and when.

Source: PROSPERO, Centre for Reviews and Dissemination, University of York, funded by NIHR. Screenshot of an earlier version of the site; the current interface may look different.
Why registration matters
CRD states four purposes for the register, and they are worth separating because they do different work for you as an author.2
| Purpose | CRD’s wording | What it changes for you |
|---|---|---|
| Avoiding wasted effort | “to help avoid unplanned duplication”2 | You find out before you invest months whether the same review is already under way. |
| Recording intent early | A listing of reviews “registered at inception”2 | The record is dated before results exist, which is what makes it evidence of intent rather than of hindsight. |
| Transparency | Enables “comparison of reported review findings with what was planned in the protocol”2 | A reader, editor or reviewer can check your published review against what you said you would do. |
| Reducing bias | “also aims to minimise the risk of bias in systematic review”2 | Outcomes and analyses fixed in advance are harder to select after the fact. |
The third purpose is the one that does the most work in practice, and it is also the one with teeth: because PRISMA 2020 requires you to describe and explain any amendments when you publish,4 the registration and the finished paper are designed to be read against each other. That only helps you if the registration was specific in the first place.
Who registers, and what they take on
Registration is not an administrative formality delegated downwards. CRD’s guidance names a single accountable person: the named contact “acts as the guarantor for the accuracy of the information presented in the register record”, and “should be the lead reviewer or a representative of the review team”.2
That person also carries the obligation forward. They are “responsible for submitting details of any amendments while the review is ongoing”, and registrants more generally are “accountable for the accuracy and updating of information submitted”.2 Registration is therefore a continuing duty for the duration of the review, not a one-off submission.
If your review changes enough that you are unsure whether it is still the same review, CRD’s instruction is unambiguous: “If in doubt, a new record for a new review should be created.”2
What can and cannot be registered
Most rejections are eligibility failures, not quality failures. CRD’s published guidance note states that PROSPERO “includes details of any ongoing systematic review that has a health related outcome in the broadest sense”, and that registered protocols “can include studies of any design”.2 CRD describes the register itself as “an international database of prospectively registered systematic reviews in health and social care”.2
The distinction that trips people up is this: the gate is the type of review and whether it has a health-related outcome — not the design of the studies you intend to include. Randomised trials, cohort and case-control studies, diagnostic accuracy studies and qualitative studies can all sit inside a registrable review. A scoping review of randomised trials is still ineligible; a systematic review of qualitative studies with a health-related outcome is not. The exclusions are specific, and they are where submissions come unstuck:
A second exclusion catches Cochrane authors in particular: “Cochrane protocols should not be registered separately with PROSPERO”,2 because new Cochrane protocols are loaded into the register automatically. Registering one by hand creates a duplicate record rather than a second registration.
PROSPERO’s scope has widened since launch, which is why older advice circulates and misleads. The register’s own design paper, published in 2012, describes a markedly narrower remit than the guidance that followed.1
| Question | 2012 design paper1 | 2016 guidance note2 |
|---|---|---|
| Subject remit | “restricted to systematic reviews of the effects of interventions” | “any health related outcome in the broadest sense” |
| Study designs | Effects of interventions | “studies of any design” |
| Reviews of reviews | “not being included at this time” | “will be accepted for registration” |
| Scoping / literature reviews | Scoping reviews “not being included at this time” | Explicitly “not eligible” |
| Reviews of animal studies | “excluded” | Not repeated in this guidance |
Two of those rows have moved in the author’s favour: reviews of reviews and non-intervention health questions are now registrable. The animal-studies row is the one to treat with care — the 2012 exclusion is not restated in the 2016 guidance, and the two documents cannot settle the current position between them. Check the live criteria before relying on it either way.
If PROSPERO will not take your review
The exclusion of scoping and literature reviews is a scope rule, not a verdict on quality — and it leaves a practical question unanswered: where does a scoping review protocol go instead?
PRISMA-ScR, the reporting extension written for scoping reviews, answers it directly. A scoping review protocol “should be developed a priori” and “should ideally be registered”, for example with the Open Science Framework.6
The two review types are doing different jobs. Scoping reviews “follow a systematic approach to map evidence on a topic”, and where a systematic review answers a clearly defined question, a scoping review addresses a much broader one. PRISMA-ScR exists because scoping reviews “should have different essential reporting items from systematic reviews”, differing in matters such as risk-of-bias assessment and meta-analysis.6 Its checklist carries 20 essential reporting items and 2 optional ones.6
If the work is a scoping review, calling it systematic to get a CRD number produces a record that does not describe the study, and a study that will later be reported against the wrong checklist. Register it where it belongs and report it against PRISMA-ScR.6
When to register: two different deadlines
This is the most commonly garbled point in secondary advice, because CRD’s guidance sets out two thresholds that do different jobs. One is a recommendation; the other is the eligibility limit.
| Threshold | Stated as | What it governs |
|---|---|---|
| Recommended point | “Ideally reviews should be registered before screening against eligibility criteria commences”2 | Best practice. Registering here is what makes the record genuinely prospective. |
| Eligibility limit | Accepted “as long as they have not progressed beyond the completion of data extraction”2 | The outer boundary in that guidance — past it, the review is not registrable at all. |
The 2012 design paper sets the same pair, in the same order: registration “should take place before the researchers have started formal screening against inclusion criteria”, with reviews eligible “as long as they have not progressed beyond the point of completing data extraction”.1
CRD’s downloadable guidance note is dated May 2016 and predates both the register’s later policy announcements and its 2025 rebuild. The eligibility limit it records may no longer be the operative one, and CRD has publicised a tighter cut-off since. Treat the live eligibility criteria on the PROSPERO site as authoritative, and register as early as you can rather than as late as an older document appears to allow.
What a registration record contains
Registration is a structured dataset, not a free-text protocol upload. The design paper describes a required set covering “the key attributes of review design” alongside administrative detail: the review title in English, anticipated or actual start date, anticipated completion date, named contact and email, organisational affiliation, funding sources, conflicts of interest and current review status.1
On the methods side, 12 of the 15 review-methods fields are required — among them the review question, searches, condition or domain, participants, interventions, comparators, study types, primary outcomes, risk-of-bias assessment, data synthesis strategy and subgroup analyses.1 In practice the methods fields are where submissions are sent back, because each must be specific enough to be checkable later.
The review methods form: 15 fields, 12 required
The methods are entered in part 3 of the four-part registration form (review title and timescale, review team details, review methods, general information). It has 15 fields; 12 are required and marked with an asterisk.
Each field has its own box and guidance text; required fields are marked *. The record shown is an NHS-funded review of treatments for frozen shoulder.

Source: PROSPERO, Centre for Reviews and Dissemination, University of York, funded by NIHR. Screenshot of an earlier version of the form; field names and numbering can change, so check the current form before submitting.
| # | Field | Status | What to include |
|---|---|---|---|
| 1 | Review question(s) | Required | The question or objectives, usually framed with PICO; one box for each question. |
| 2 | Searches | Required | The sources to be searched and any restrictions, such as language or publication period. |
| 3 | URL to search strategy | Optional | A link to the full search strategy, or an uploaded PDF. |
| 4 | Condition or domain being studied | Required | A short description of the disease, condition or health care domain. |
| 5 | Participants/population | Required | Who will be included, and who will be excluded. |
| 6 | Intervention(s), exposure(s) | Required | The intervention or exposure being assessed. |
| 7 | Comparator(s)/control | Required | What the intervention will be compared with. |
| 8 | Types of study to be included | Required | The study designs that will be included and excluded. |
| 9 | Context | Optional | Any setting criteria, such as country, care setting or time period. |
| 10 | Primary outcome(s) | Required | The main outcome, how it is measured and when. |
| 11 | Secondary outcome(s) | Required | Other outcomes, including harms, with measures and timing. |
| 12 | Data extraction (selection and coding) | Optional | How studies are selected and data extracted, by how many reviewers, and how disagreements are resolved. |
| 13 | Risk of bias (quality) assessment | Required | The named tool and how it will be applied. |
| 14 | Strategy for data synthesis | Required | The planned analysis, effect measure, model and handling of heterogeneity. |
| 15 | Analysis of subgroups or subsets | Required | Planned subgroup analyses, stated in advance, or “none planned”. |
The three optional fields (search strategy URL, context and data extraction) are still worth completing. A linked search strategy and a stated extraction process make the review easier to reproduce and to check against the published version.
The registration is not the protocol
These are two different documents, and submitting one does not produce the other. The PROSPERO record is a structured dataset of required fields. The protocol is a full methods document, written to be followed.
PRISMA-P 2015 is the reporting guideline for the protocol,7 with a companion explanation and elaboration paper that works through each item in detail.9
PRISMA 2020 asks for both, separately. Item 24a wants “registration information for the review, including register name and registration number”; item 24b asks you to “indicate where the review protocol can be accessed, or state that a protocol was not prepared”.4 Two items, because they are two artefacts — and 24b explicitly contemplates a registered review with no protocol behind it.
How good are registrations in practice?
Registration is now routine. Whether the records are specific enough to be worth anything is a separate question, and it has been measured.
Page, Shamseer and Tricco examined a random sample of 150 records registered between April and September 2017. Only 9 of the 150 — 6% — fully pre-specified the primary outcome. In a further 44 of 150 (29%), only the outcome domain was given, without the detail that would make it checkable later.8
Registration itself is straightforward. Registering in sufficient detail to be audited against later is where most records fall short. A primary outcome stated as a domain — “pain”, “quality of life” — without the measure, the timing and the effect measure leaves you free to choose among them later, which is precisely the freedom registration is supposed to remove. The same authors recommend that records match the public protocol and that the optional timing and effect-measure fields be used properly.8
For scale, the same paper reports 26,535 registered systematic reviews as of 10 October 2017, with monthly registrations rising roughly tenfold — from 63 a month in 2012 to about 800 a month in 2017 — and notes that registration reduces “waste from unintended duplication”.8 Those figures are a snapshot from 2017; the register has grown by an order of magnitude since.3
What this looks like in practice
Three views of the material above: when the two deadlines fall, how eligibility is decided, and how specific registered records actually are.
CRD publishes a recommended point and an eligibility limit. The advice repeated almost everywhere sits between them.
What to look for. The green marker and the red marker are not the same decision. CRD recommends registering “before screening against eligibility criteria commences”, while its published eligibility limit is “not progressed beyond the completion of data extraction”.2 Aim at the green one.
Schematic. Stage spacing is illustrative; CRD has publicised a tighter cut-off since the 2016 guidance, so check the live criteria.
The four questions CRD’s rules reduce to, in the order worth asking them.
What to look for. Three of the four exits are scope rules rather than quality judgements.2 The scoping review exit is not a dead end: PRISMA-ScR says such a protocol “should ideally be registered”, for example with the Open Science Framework.6
Each square is one of 150 PROSPERO records sampled at random from April to September 2017.
This one is real data, not a schematic. Page, Shamseer and Tricco sampled 150 records registered between April and September 2017. Nine of them — 6% — fully pre-specified the primary outcome; a further 44 (29%) gave only the outcome domain.8 The remaining 97 follow by subtraction. Registration itself is straightforward; registering in sufficient detail to be checked against later is where most records fall short.
Figures are from a 2017 sample and may not describe records registered since.
Common misunderstandings
Five expectations are worth dismantling, because each produces a predictable and avoidable problem.
“Registration means my methods have been approved.” It does not. CRD checks eligibility, clarity and whether required fields are complete — and the design paper is explicit about the limit of that: “The checks made do not constitute peer review or imply approval of the methods proposed for the review being registered.”1 A registered record carries no endorsement of your methods.
“A CRD number helps the review get published.” Registration records a plan. Nothing follows from it about whether a journal will accept the finished review.
“PROSPERO is really for Cochrane reviews.” Any eligible health-related systematic review can be registered. The reverse is in fact the rule: Cochrane protocols specifically should not be submitted by hand, because they are loaded automatically.2
“My review is close enough to a systematic review to qualify.” Eligibility is categorical, not a matter of degree. Scoping reviews and literature reviews “are not eligible for inclusion in PROSPERO”,2 however systematically they were conducted.
“Once it is registered, it is fixed.” The opposite: the record is designed to be amended, and the named contact is “responsible for submitting details of any amendments while the review is ongoing”.2 What you cannot do is amend it invisibly, because every change is dated and public.
Amendments and the public version history
A registration is not frozen, but neither is it quietly editable. CRD’s guidance is that “all submitted edits and changes to a PROSPERO record will be recorded, dated”, that “the most recent version will appear with previous versions accessible from dated links”, and that changes require brief details which themselves “appear in the public record”.2
This is the mechanism that gives registration its evidential value: a reader can see what you planned, what you changed and when. It also means deviations are better declared than discovered. Registration is free of charge, and registrants are “accountable for the accuracy and updating of information submitted”.2
PROSPERO and PRISMA 2020
The two instruments are often conflated. PROSPERO records what you planned; PRISMA 2020 governs how you report what you did. PRISMA 2020’s checklist carries a “Registration and protocol” section whose three items are reporting requirements:4
- 24a — “Provide registration information for the review, including register name and registration number”
- 24b — “Indicate where the review protocol can be accessed, or state that a protocol was not prepared.”
- 24c — “Describe and explain any amendments to information provided at registration or in the protocol.”
Items 24a–24c require you to report your registration status and any amendments. They are reporting items in a reporting guideline — they do not themselves make registration mandatory, and 24b expressly accommodates a review for which no protocol was prepared. Individual journals may impose their own requirements; PRISMA does not.
Item 24c is the one that connects back to the version history: if you amended the registration, the published review has to say so and explain why. A dated public trail makes that straightforward, which is a further argument for registering early and amending openly rather than registering late with everything already settled.
What changed in 2025
PROSPERO was rebuilt during 2025. CRD announced on 7 March 2025 that the register “has been completely redesigned”, with new features “to improve the registration process” and “advanced automatic processing” to shorten the time records take to process.5 Two consequences are worth planning around.
First, publication is now automatic once checks and author approval are complete: “all submissions will be published automatically after checking and approval by all record authors.”5 Every named author has to approve the record, so a co-author who does not act will hold up publication.
Second, CRD signalled in January 2025 that the new version would flag “duplicate or very similar topics at submission”, explicitly to reduce research waste.3 Searching the register for near-identical reviews before you submit is therefore no longer merely good manners — it anticipates a check the platform now makes.
Take-home message
PROSPERO records a plan and timestamps every later change to it; it certifies nothing about your methods. Register before screening begins rather than at the latest moment an older document appears to permit, and check eligibility first — scoping and literature reviews are turned away outright.
How to apply this to your work
- Check eligibility before you write anything. If it is a scoping or literature review, PROSPERO is the wrong register and rewriting the title will not change that.2
- Search the register for overlapping reviews first. With records running at roughly 215 a day, and duplicate-topic flagging now built into submission, an unchecked overlap is likely to surface anyway.3
- Register before screening against eligibility criteria begins. That is CRD’s recommended point, not the outer limit — aim at the recommendation.2
- Draft the 12 required methods fields to be auditable. Each should be specific enough that a reader can later check what you did against what you said.1
- Line up co-author approval in advance. Publication now waits on approval by all record authors, so an unresponsive co-author stalls the record.5
- Amend openly, and report the amendments. The version history is dated and public; PRISMA item 24c requires you to describe and explain the changes.4
Common questions
Does registering on PROSPERO mean my methods have been checked?
No. CRD checks eligibility, clarity and completeness of required fields, and the register’s design paper states plainly that these checks “do not constitute peer review or imply approval of the methods proposed for the review being registered”.1 A CRD number records that you registered a plan, nothing more.
Can I register a scoping review?
Not on PROSPERO. CRD’s guidance states that “scoping reviews and literature reviews are not eligible for inclusion in PROSPERO”.2 This is a scope rule, not a quality judgement, and it is one of the more common reasons a submission is returned.
I have already started screening. Is it too late to register?
Probably not, but you are past the recommended point. CRD’s published guidance accepts reviews “as long as they have not progressed beyond the completion of data extraction”, while recommending registration “before screening against eligibility criteria commences”.2 Because that guidance note dates from May 2016 and a tighter cut-off has been publicised since, check the live eligibility criteria rather than relying on the older limit.
Do journals require PROSPERO registration?
PRISMA 2020 does not. Its items 24a–24c are reporting requirements — you must report your registration information and any amendments, and 24b expressly allows for a review where no protocol was prepared.4 Individual journals may set their own policies, so check the specific instructions for authors.
Common mistakes when registering
Most of these are settled in the hour you spend filling in the form, and most are expensive to undo.
| Mistake | Stage | What it costs | What to do instead |
|---|---|---|---|
| Submitting a scoping or literature review | Before registering | Rejection on eligibility, whatever the quality of the work2 | Register a scoping review protocol elsewhere — PRISMA-ScR points to the Open Science Framework6 |
| Registering a Cochrane protocol by hand | Before registering | A duplicate record, since new Cochrane protocols are loaded automatically2 | Let the automatic feed do it |
| Not searching the register first | Before registering | Months spent duplicating a review already under way; duplicate topics are now flagged at submission3 | Search the register before you write the submission8 |
| Waiting until the outer limit to register | Timing | A record that is barely prospective, and advice that may be out of date given CRD has publicised a tighter cut-off | Register before screening against eligibility criteria commences — CRD’s recommended point, not its limit2 |
| Stating the primary outcome as a bare domain | Filling in the form | The commonest failure in the register: only 6% of sampled records fully pre-specified the primary outcome8 | Give the measure, the timing and the effect measure, not just “pain” or “mortality”8 |
| Treating the record as the protocol | Filling in the form | No methods document to follow, and nothing to answer PRISMA item 24b with4 | Write a protocol to PRISMA-P as well as registering7 |
| Naming a contact who is not accountable | Filling in the form | Amendments go unsubmitted; the named contact is “the guarantor for the accuracy of the information”2 | Name the lead reviewer or a representative of the review team2 |
| Letting the record go stale | During the review | Registrants are “accountable for the accuracy and updating of information submitted”2, and a stale record undermines the comparison registration exists to enable | Submit amendments as the review develops; every change is dated and public2 |
| Changing the review so much it is a different review | During the review | A record that no longer describes the study | CRD’s instruction is unambiguous: “If in doubt, a new record for a new review should be created”2 |
| Not lining up co-author approval | Submission | Publication stalls — all record authors must approve before a submission is published5 | Warn every named author that an approval request is coming5 |
| Amending quietly and not reporting it | Write-up | PRISMA item 24c requires you to “describe and explain any amendments”4, and the dated public history makes an unexplained change visible | Report the amendments and the reasons in the paper |
Declared interest: Pubrica provides systematic review and protocol development support commercially. Nothing on this page should be read as evidence that registration, or paid assistance with it, improves the likelihood of publication — no claim of that kind is made here, because we are not aware of independent evidence supporting one.
References and sources
- Booth A, Clarke M, Dooley G, Ghersi D, Moher D, Petticrew M, Stewart L. The nuts and bolts of PROSPERO: an international prospective register of systematic reviews. Systematic Reviews. 2012;1:2. doi:10.1186/2046-4053-1-2 read 19 October 2026.View on Systematic Reviews
- Centre for Reviews and Dissemination. Guidance notes for registering a systematic review protocol with PROSPERO. University of York; May 2016. read 19 October 2026.View on CRD, University of York
- Centre for Reviews and Dissemination. Number of systematic reviews registered in PROSPERO passes 300,000: how the numbers have changed over time. University of York; 6 January 2025. read 19 October 2026.View on University of York
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. Checklist items 24a–24c, “Registration and protocol”. read 19 October 2026, from the PRISMA 2020 checklist as distributed by the EQUATOR Network.View on EQUATOR Network
- Centre for Reviews and Dissemination. PROSPERO update. University of York; 7 March 2025. read 19 October 2026.View on University of York
- Tricco AC, Lillie E, Zarin W, O’Brien KK, Colquhoun H, Levac D, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med. 2018;169(7):467–473. doi:10.7326/M18-0850 read 20 October 2026.View on the PRISMA statement site
- Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4(1):1. doi:10.1186/2046-4053-4-1 read 20 October 2026.View on the PRISMA statement site
- Page MJ, Shamseer L, Tricco AC. Registration of systematic reviews in PROSPERO: 30,000 records and counting. Syst Rev. 2018;7:32. doi:10.1186/s13643-018-0699-4 read 20 October 2026.View on Systematic Reviews
- Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015;349:g7647. doi:10.1136/bmj.g7647 read 20 October 2026.View on the PRISMA statement site
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Confirm before you open the registration form.
[ ] It is a systematic review, not a scoping or literature review [ ] It has a health-related outcome [ ] It is not a Cochrane protocol (those are loaded automatically) [ ] No equivalent review already registered on the same question [ ] Screening against eligibility criteria has not yet begun [ ] Live eligibility criteria checked on the PROSPERO site [ ] Named contact agreed — lead reviewer or team representative [ ] All 12 required methods fields drafted and specific [ ] Search strategy written down and reusable [ ] Every record author ready to approve
We screen eligibility before you draft, check the register for an overlapping review, and write the twelve required methods fields so they can still be audited against your finished paper.