Best screening tools for a systematic review

Screening tools for a systematic review do the same narrow job in different shapes: they take a pile of titles and abstracts and help two people decide what stays. They do not all do the job the same way, and several of them are built to keep going into full text, extraction, or both.

This page is a list, not a league table. There is no score, no star rating, and no claim about recall, accuracy, or hours saved. Each note below was checked on that vendor’s own site. Verified Oct 2026.

Disclosure: we make Study Screener. We rank it where it fits and name the tools that suit some teams better.

If you are still deciding what screening is, start with what screening means in a systematic review and the beginner’s step-by-step guide. For a wider catalogue beyond this list, use SystematicReviewTools.app.

How to use this list

For each tool: what the vendor’s site actually describes, who it is best for, and who it is not ideal for. “Best for” means a job the product is built around. It is not a rank against the other names.

A tool can be the right pick for one team and a poor pick for the next. Covidence, Nested Knowledge, EPPI-Reviewer, and DistillerSR are clearly better than a screening-only tool when the review has to continue into extraction or a coded synthesis. ASReview is clearly better when you want open-source active learning on infrastructure you control. Those are different jobs. Study Screener is described at the end as a title and abstract tool, not as a winner.

Rayyan

Rayyan describes a web and mobile review that runs from import through title and abstract screening, and its help centre now documents later stages as well.

Checked on Rayyan’s help pages:

• Import includes EndNote .enw, RIS, BibTeX, CSV, PubMed NBIB and XML, Web of Science CIW, and PDF. Supported formats (help updated 22 July 2026).
• Duplicate detection flags possible duplicates for a manual decision or the Auto-Resolver. Duplicates (help updated 17 May 2026).
• Blind Mode is on by default. Reviewers see only their own decisions until it is turned off. The conflict filter needs Blind Mode off. Blind Mode (help updated 14 May 2026).
• Screening decisions are Include, Maybe, and Exclude. How to screen. Conflict handling depends on a per-review method. Consensus is the default: an Include versus Exclude disagreement stays unresolved until the decisions match, and Maybe alone does not open a conflict. Majority can resolve a clear majority on its own. Conflicts (help updated 5 October 2026).
• Help also documents full-text screening, data extraction, and risk of bias as later stages, plus a PRISMA 2020 flow diagram that fills some counts from the review and leaves others for you to type. Downloads are PDF, PNG, SVG, and interactive HTML, on Essential, Advanced, and institutional plans. Full text, extraction, PRISMA diagram (help updated 21 July 2026).
• Exports from screening include RIS, BibTeX, EndNote, and CSV, with an option to include decisions. Export.
• The product site still markets a mobile app and AI-assisted prioritisation while you screen.

Best for. Teams who want collaborative title and abstract screening with in-app deduplication, a mobile app, and the option to continue into full text and extraction in the same review.

Not ideal for. Groups who want open-source software they can inspect and run themselves, or a coding tool aimed at complex evidence maps. A side-by-side with Study Screener is in Study Screener vs Rayyan.

Covidence

Covidence presents one hosted workflow: import, screening, extraction, quality assessment, and export. The homepage says duplicates are removed before screening, dual screening is independent and blinded, and full-text exclusion reasons feed the PRISMA record.

Checked on Covidence’s help:

• Imports accept EndNote XML, PubMed text, and RIS. Duplicates are removed on import and shown on the PRISMA chart. Import studies.
• Title and abstract votes are Yes, Maybe, or No. A new review defaults to dual screening: two different people must vote, and voting is blinded until each person has voted. Title and abstract screening (help updated 16 December 2025).
• That same article says the default order “Most relevant” uses machine-learning predictions. You can switch the order to author, title, or most recent.
• Team settings can limit who screens, who reviews full text, and who extracts. Team settings.
• Extraction is a separate stage, with two template styles (Extraction 1 and Extraction 2), including dual extraction and consensus on Extraction 1. Which extraction to use.
• The PRISMA flowchart updates as the review moves and downloads as a DOCX file. Some downloaded fields, including studies not retrieved, still need a manual check. Export PRISMA.

Best for. Teams, including many library-supported groups, who want screening, full text, extraction, and quality assessment in one hosted review, with a PRISMA diagram that updates from those stages.

Not ideal for. A short student project that only needs title and abstract decisions and a file export. Covidence is a fuller review system than that job requires.

ASReview

ASReview is open-source screening software coordinated at Utrecht University. The project site describes AI screening, simulation and benchmarking, and crowdscreening.

Checked on the project docs:

• You label a title and abstract as relevant or irrelevant. The software trains from those labels and reorders the remaining records so records it treats as more relevant come next. Screening.
• RIS (.ris or .txt) is a first-class format, including files from Zotero and EndNote. Labels can travel in the RIS notes field. Data format.
• You can export RIS, CSV, TSV, or Excel. Progress and export.
• The information-specialist page says there is no subscription fee, and that you need a computer with Python or a server your organisation provides. For information specialists.

Best for. Methods-confident reviewers who want open-source active learning, a simulation mode, and data that stays on infrastructure they control.

Not ideal for. A team that wants a hosted, blinded dual-screening inbox with a third person resolving conflicts, and no install. ASReview is the better fit when that open-source control matters more than a turnkey review site. It is not an extraction suite.

DistillerSR

DistillerSR describes an AI-enabled literature review platform aimed at organisations that run many reviews and want a configurable, auditable workflow. The product page is written for life-sciences companies as well as research groups.

Checked on that page, ignoring any time-saved or adoption figures:

• Workflows can be configured and reused across reviews.
• AI re-ranking uses ongoing include and exclude decisions to move unscreened references up the queue. Separate classifiers can label references by categories such as study design or PICO element.
• Extraction forms can suggest answers. The page says a person accepts or rejects suggested values, and that suggestions point back to the source.
• An audit trail and version history are part of the product description, along with reporting and integrations to libraries and business-intelligence tools.

Best for. Organisations that need a configurable literature-review pipeline, reuse of extracted data, and an audit trail across many projects.

Not ideal for. A single student review or a team that only needs title and abstract screening. DistillerSR is a heavier platform than that. Study Screener is not trying to be this product.

EPPI-Reviewer

EPPI-Reviewer is developed by the EPPI Centre at the UCL Institute of Education. The software page describes a browser app (EPPI-Reviewer 6) for systematic reviews, meta-analyses, narrative reviews, and meta-ethnographies, including very large reviews.

Checked on EPPI Centre pages:

• Automation tools include deduplication, priority screening, and pre-built study-type classifiers. Automation tools.
• Priority screening learns from title and abstract include and exclude decisions and orders the remaining records. The Centre’s machine-learning note says the tool is designed for title and abstract screening and recommends that reviewers still screen every record. Machine learning in EPPI-Reviewer.
• Coding tools cover screening and data extraction. The EPPI-Reviewer 6 meta-analysis guide shows outcome, intervention, and comparison codes used to set up extraction for synthesis. Setting up a meta-analysis in EPPI-Reviewer 6.

Best for. Review teams, especially in social science and public health, who want flexible coding from screening through extraction and on toward meta-analysis, inside the EPPI Centre’s tool.

Not ideal for. Someone who wants a short, modern onboarding for a first title and abstract pass. The coding model is powerful and it is more to learn. EPPI-Reviewer is the better fit when that coding depth is the point.

Abstrackr

Abstrackr is still available, at a new address. The older Brown University host (abstrackr.cebm.brown.edu) showed a notice dated 23 April 2025: that copy would be deprecated and would stop working by the end of 2025, and it pointed people to abstrackr.com. The new site was online when checked in October 2026. Treat the Brown address as retired.

Checked on abstrackr.com:

• The site describes a free citation-screening tool with single-reviewer and double-screening modes.
• Blinding is described as on by default. Disagreements and Maybe votes are described as conflicts a leader can resolve side by side.
• After a set of labels, a ranking queue orders remaining citations. The site also describes a pilot round, assignments, and CSV or RIS export.
• A live PRISMA numbers table is part of the product description. The site also describes a stopping analysis. Whether you stop before every record is screened is a protocol decision; do not treat a vendor screen as permission to stop early.

Best for. Teams who want a free hosted screening tool with blinding, conflicts, and a ranked queue, and who are comfortable on the current abstrackr.com site rather than the old Brown host.

Not ideal for. Reviews that need extraction, quality assessment, or a long institutional contract. Confirm the new site against your protocol before you move a live project off another tool.

PICO Portal

PICO Portal describes a hosted literature-review platform covering title and abstract screening, with priority screening, deduplication, and later extraction and quality assessment. The marketing pages quote efficiency figures; those figures are not repeated here.

Checked on the product docs and features page:

• You can import from reference libraries or a CSV template. The about page describes deduplication and keyword highlighting on titles and abstracts, then prioritisation so references treated as more relevant appear first. About PICO Portal.
• Workflows can differ by stage. Dual independent review means two reviewers screen blind, and a judicator resolves disagreements. You can also choose a single review, a principal-reviewer rule, or a calibration step, separately for abstract screening, full text, and extraction. Create a project.
• The features page says you can download a PRISMA 2020 diagram and use AI-assisted quality assessment and data extraction. Features.

Best for. Teams who want hosted priority screening, highlighted terms, a named judicator, and a path into full text and extraction in the same project.

Not ideal for. Reviewers who need the software to be open source. Priority screening is part of the product and can be left off: the features page describes the AI pieces as optional. Features. PICO Portal is the better fit when that hosted path into extraction is what you actually need.

Nested Knowledge

Nested Knowledge documents a connected workflow it calls AutoLit: search, screening, tagging, extraction, and critical appraisal, with synthesis views after tagging. Its docs were updated through 2026.

Checked on those docs:

• You can run searches against several indexes, or import RIS and NBIB files. Deduplication runs on import, with a duplicate-review step. Literature search, import files.
• Dual screening means two people screen each record and an administrator adjudicates. Agreed decisions can be auto-adjudicated. The same page documents kappa statistics in the activity view after dual screening. Dual screening (updated 3 August 2026).
• Dual two-pass screening does that twice: abstract, then full text. Dual two-pass.
• Included records move on to tagging, which the docs describe as the extraction step that feeds qualitative synthesis. An optional AI screener can occupy one dual-screening seat; a person occupies the other, and an administrator sets the final decision.

Best for. Teams who want screening, tagging, extraction, and a synthesis view in one nest, including dual review with adjudication and a documented kappa view.

Not ideal for. A reviewer who only needs a title and abstract decision log and a RIS export. Nested Knowledge is the better fit when the nest itself is the review workspace. It is more product than a screening queue.

Study Screener

Study Screener is a focused title and abstract screening tool. It is not an extraction suite. After you have included studies, build the evidence table in Evidence Table Builder.

What the product does, as implemented in this project:

• Import a RIS file, including a RIS exported from Zotero or EndNote, or connect a Zotero collection with an API key. PubMed MEDLINE .txt exports are accepted too. Steps: Import from Zotero, EndNote, or a RIS file.
• Dual screening with a conflict list after the project is opened, and Cohen’s kappa on overlapping decisions. How the agreement statistic is meant to be read: Dual screening, conflicts, and Cohen’s kappa.
• AI-assisted prioritisation as a separate job, not a re-rank of the manual queue while you click. The job labels records and attaches a confidence score so you can review uncertain records first. How that differs from a fully manual dual review: AI screening vs manual screening.

Best for. Teams whose immediate job is title and abstract screening: RIS or Zotero in, blinded dual review, conflicts, kappa, and an optional AI pass that orders records by confidence. A closer look at one alternative is Study Screener vs Rayyan.

Not ideal for. Anyone who needs full-text PDF screening, extraction forms, or risk-of-bias modules in the same subscription. Covidence, Nested Knowledge, EPPI-Reviewer, DistillerSR, and Rayyan’s later stages are the better fit for that wider job. Use Study Screener for the screening step, then Evidence Table Builder for extraction.

Which tool to open first

None of these rows is a score. A library subscription, a protocol that forbids stopping early, or a requirement to keep data on your own server can overturn the row. SystematicReviewTools.app is the wider map if the tool you need is not a screener at all.

Related reading

• What screening is in a systematic review
• Paper screening: a beginner’s step-by-step guide
• Study Screener vs Rayyan
• Dual screening, conflicts, and Cohen’s kappa
• Import from Zotero, EndNote, or a RIS file
• AI screening vs manual screening

---

Vendor claims verified Oct 2026 on the pages linked above. Product names and plan details change. Re-check the vendor site before you commit a review to it. Methods context: PRISMA 2020 and Cochrane Handbook, Chapter 4.