Which systematic review tool should you use?
We make one of these. We will still tell you when one of the others is the better answer, because you will find out either way and it is cheaper for both of us if you find out now.
Most review teams end up using two or three tools at once — one for deduplication, one for screening, one for writing up. That is not a failure of discipline. It is what happens when no single tool was built for the kind of review you are doing.
Start here: what kind of review is it?
Use Covidence.
It is the default in health sciences for good reasons: Cochrane association, a complete workflow, risk-of-bias tooling, and your supervisor and examiners already know what its output looks like. If your institution pays for it, it is free at the point of use to you and does more than we do.
$339/year for one review if your library does not cover it. Annual only, no student rate.
Use Rayyan or ASReview.
Rayyan's free tier handles three reviews with two reviewers and is genuinely good at collaborative screening. ASReview is free, open source, and has machine-learning prioritisation that can materially cut the screening burden on a large set — which we do not have.
Rayyan free, or roughly $5–8 per seat per month. ASReview free. Check whether the free tier makes your project public before you rely on it.
Use EPPI-Reviewer.
Built at UCL's EPPI-Centre for exactly these fields, with text mining and qualitative synthesis, and free for Cochrane and Campbell reviews. If your review is one of those, this is the strongest free option in the list.
Use CADIMA.
Free, and the de facto default in that field. Being the tool your reviewers already recognise is worth more than a feature comparison.
Use Elicit.
A structured review workflow over 138 million papers with model-assisted screening, and it is fast. Two things to know before you commit. Reporting standards are moving toward requiring disclosure of any AI used in screening — the tool, its threshold, your checking strategy, agreement rates — so plan that into your protocol rather than discovering it at write-up. And you will be asked, in a viva or by a reviewer, how you decided; "the model screened and I checked a sample" is a defensible answer but it is one you have to defend.
$12/month Plus, $49/month Pro, $79 per seat on Team.
That is the one we built for.
Every tool above except EPPI-Reviewer and CADIMA was built around the clinical evidence hierarchy, where an interview study sits near the bottom. If your evidence is qualitative, mixed-method or interpretive, you are using a clinical instrument on a question it was not designed for, and the appraisal step will keep telling you your best sources are weak.
What we do differently, and it is a short list: sources are appraised within their own paradigm, so an ethnography is never ranked beneath a trial. You screen every record, logged with a reason in a record that cannot be edited afterwards — and you can export a statement of exactly how the review was conducted. Your supervisor and second screener are free, on every plan. Your work is private by default, which matters more than it sounds for an unpublished thesis.
What we do not have, plainly: no machine-learning screening prioritisation, no risk-of-bias tooling, no full-text handling, no reference-manager integration yet. If any of those is the thing you need most, pick from the list above instead.
The free tools here work regardless of what you choose
Seven of them, in your browser, no account, permanently free — and they do not care which platform you run your review in. The PRISMA flow diagram generator, a search string builder covering five frameworks including the non-clinical ones, a search log builder, a question health check, a journal credibility check, a reference formatter, and one-page framework references.
Where these figures came from
Prices and features are what each product publishes, checked August 2026. We have not bought any of them, so this is their advertising, not our measurement — and a comparison page written by a competitor deserves that caveat stated rather than buried. If something here is out of date or unfair, tell us and it will be corrected.