Rayyan is a web-based screening tool for systematic reviews that lets two or more reviewers work through imported records independently, hiding each other’s decisions, while a machine-learning model learns from those decisions to push the most likely-relevant studies toward the top of the queue. It is free to start and run a review with its core features, with paid tiers for heavier or team use, which is why it is a common choice for student projects, unfunded reviews, and groups that need solid blinded screening without an institutional budget.
What Rayyan does in a review
Rayyan covers the screening stages of a review rather than the whole pipeline. You import the deduplicated search results, screen on title and abstract with the team blinded to one another, label and filter records as you go, and export the included set for full-text work and extraction elsewhere. It sits in the same category as other dedicated systematic review screening tools, and it is most often compared with Covidence; the practical difference is that Rayyan concentrates on fast, assisted screening, whereas a tool like Covidence also builds in guided extraction and risk-of-bias steps. Understanding that scope keeps expectations right: Rayyan is excellent for the title and abstract screening pass and capable at full text, but the synthesis happens in other tools.
Blinded dual screening
The core methodological feature is blinded dual screening. Rayyan can hide each reviewer’s include and exclude decisions from the others until the blind is lifted, so one person’s judgement does not anchor the rest of the team. When the blind comes off, Rayyan shows where reviewers agreed and where they conflicted, and the team resolves the disagreements. This is the same two-reviewer discipline that separates a defensible screen from a single-person sift, and the decision data can feed an agreement statistic if you take it into a Cohen’s kappa calculator. How many reviewers each stage needs is a design choice in its own right, discussed in our guide on how many reviewers to use for screening.
Labels, filters, and organising the queue
Rayyan gives reviewers flexible labels and filters to organise a large record set. You can tag records with custom labels, for example a reason a study looks excludable or a topic it belongs to, and then filter the queue to focus on one slice at a time. Rayyan also detects likely duplicates and flags terms it has spotted across records, which helps a reviewer move quickly through a long list. These tools speed the work but do not replace a clean import: feeding Rayyan a properly deduplicated set of search results first means the labels and filters are working on unique studies rather than repeated ones.
Machine-learning relevance ranking
Rayyan’s best-known feature is relevance ranking. As reviewers include and exclude records, a model learns the characteristics of the studies the team is keeping and reorders the unscreened queue so that the records most likely to be relevant rise to the top. The practical benefit is that reviewers often encounter the bulk of the includable studies early in the queue, which can make a large screen feel faster and helps a team gauge when the relevant material is thinning out. The important caveat is that this ranking is an aid to the order of work, not a licence to stop early: a systematic review still requires that every record be screened by humans, because the model can be wrong about an individual study. Used as a prioritiser rather than a decider, the ranking is a genuine time-saver.
Importing and exporting
Rayyan imports the standard formats databases produce, including RIS and other tagged exports, so the file that leaves your reference manager loads directly into a new review. The natural sequence is to run the search, gather everything in a reference manager, deduplicate, then import the clean set into Rayyan; our companion guides to screening with Covidence describe the equivalent flow for that platform. When screening is done, Rayyan exports the decisions and the included records so they can move into full-text review and extraction, and the running counts at each stage feed the figures a PRISMA flow diagram needs. Keeping the import file and the exported decisions is part of an auditable trail.
Resolving conflicts and keeping an audit trail
The moment that decides whether a Rayyan screen will survive peer review is conflict resolution. After the blind is lifted, the records split into agreed includes, agreed excludes, and conflicts where the two reviewers disagreed, and the team has to settle every conflict by a pre-agreed rule rather than by whoever feels strongest on the day. A common and defensible approach is for the two reviewers to discuss each conflict and, where they cannot agree, to bring in a third reviewer as an arbiter. Rayyan keeps the include and exclude marks visible through this process, which is what lets you reconstruct later how each contested record was settled. That reconstruction is the point: a systematic review is meant to be auditable, so capture the number of conflicts and how they were resolved, exactly the discipline set out in our guide to resolving screening conflicts. The same care carries into the next stage, where the included reports are read in full and a second set of exclusions is recorded with reasons during full-text screening, and those reason counts are what populate the lower half of the flow diagram.
When teams choose Rayyan over Covidence
The decision usually comes down to budget and scope. Teams choose Rayyan when cost is the deciding factor, since its core screening is free, when they only need screening rather than built-in extraction, and when they value the relevance ranking for a very large record set. They lean toward Covidence when they want a single guided environment that carries the review through extraction and risk-of-bias assessment as well, and when an institutional subscription already covers it. Many groups happily use both: Rayyan for the screening pass, then a separate extraction step elsewhere. If you are still mapping out the project as a whole, our guide on how to write a systematic review shows where the screening stage fits among everything else, so you can pick the tool that matches the rest of your plan rather than the other way round.