Disclosure first: Covexe is our product. It earns its spot on this list, but every other entry is judged on its merits, and for several readers one of the others is the better choice. Prices and access terms are as of 2026.
R with the metafor package
metafor is the reference implementation of meta-analysis. It is a free R package that covers pairwise synthesis in more depth than anything else here: every common effect measure, fixed and random effects with a long list of tau-squared estimators, meta-regression, multilevel and multivariate models, and publication bias diagnostics. Covexe validates its meta-analysis output against it.
The limit is the obvious one: you write code. If you have never used R, budget real time for the learning curve. And metafor does nothing for the review around the analysis: no screening, no PRISMA diagram, no risk of bias tables. Pick it if you are comfortable scripting, need unusual models, or want an analysis a statistician can audit line by line.
RevMan
RevMan Web is Cochrane's own review software, now hosted in the browser (the desktop RevMan 5 was discontinued). It is free for Cochrane authors working on a Cochrane review and for users in Research4Life countries. Everyone else needs a paid subscription, with discounted academic rates and student licenses around 60 GBP.
Its strength is the workflow: review structure, risk of bias tables, and forest plots in exactly the format Cochrane expects. Its limit is the same fact. Outside a Cochrane review you pay for it and inherit conventions built for someone else's process, a gap our RevMan alternative page covers in detail. Pick RevMan if you are writing a Cochrane review; in that case it is barely a choice, you will be using it anyway.
JASP and jamovi
Both are free desktop statistics programs with a point-and-click interface and a meta-analysis module. If you already run your general statistics in one of them, the module gets you pooled estimates, forest plots, and heterogeneity output without writing code. JASP is also the rare free GUI that runs Bayesian meta-analysis, worth knowing if a reviewer asks for one.
Neither has a systematic review pipeline: no reference screening, no PRISMA diagram, no risk of bias workflow. You arrive with extracted effect sizes and leave with an analysis. They are also desktop installs, which matters if you move between machines. Pick one if meta-analysis is a small part of a stats workflow you already run there.
Covexe (our tool)
Covexe is a free browser platform that covers the review as well as the analysis: PubMed and Europe PMC search and import, deduplication, blinded dual screening with Cohen's kappa, full-text review, data extraction, risk of bias, GRADE, and a PRISMA 2020 flow diagram counted from your screening decisions. The meta-analysis engine is validated against metafor and the network meta-analysis against netmeta, and every meta-analysis comes with downloadable R replication code, so a reviewer can check the numbers independently.
Limits, honestly stated: the AI-assisted features (screening suggestions, extraction drafts) are metered on the free tier unless you bring your own API key, and while the stats suite is broad, it does not cover everything (no mixed models, no factor analysis). Pick Covexe if you want screening, appraisal, and analysis in one free place without installing anything.
MetaInsight
MetaInsight is a free web app for network meta-analysis, built on the netmeta R package. It handles the essentials in the browser: network plots, league tables, treatment ranking, and inconsistency checks, no code required.
It does one thing. There is no pairwise workflow around it and no review pipeline; you bring a finished dataset of treatment comparisons. Pick it if your only need is a network meta-analysis and your data is already extracted.
Quick browser calculators
Tools like meta-mar and MetaAnalysisOnline form a class of their own: paste effect sizes, get a pooled estimate and a forest plot. For a one-off check, a homework problem, or a quick look at whether five trials point the same way, they are genuinely useful and free.
They are thin on everything around the estimate: no project that persists, no screening, no reporting workflow. Treat them as calculators, not review software.
Meta-Essentials
Meta-Essentials is a free set of Excel workbooks from Erasmus University Rotterdam. If your world is Excel and your review is small, it produces pooled estimates, forest plots, and heterogeneity statistics inside a spreadsheet you already know, with good documentation.
The trade-offs are the spreadsheet's: manual data handling, no screening or reporting pipeline, and it stops at pairwise analysis. Pick it if staying in Excel outweighs everything else.
Comprehensive Meta-Analysis is not free
CMA is a capable desktop package, but it is not free: academic subscriptions start around 295 USD per year as of 2026, corporate rates are higher, and it runs on Windows only. That puts it outside this list. If you are weighing it against the free options, our CMA alternative page makes the direct comparison.
How to choose
- Writing a Cochrane review: RevMan. It is free for you and the workflow is built for exactly that.
- Comfortable with code, or need unusual models: R with metafor.
- Already doing your statistics in JASP or jamovi: use its meta-analysis module.
- Network meta-analysis only, data already extracted: MetaInsight.
- One quick pooled estimate to sanity-check: a browser calculator.
- Committed to Excel: Meta-Essentials.
- A full review from search to PRISMA diagram, free and in the browser: Covexe.
There is no single winner. The right pick depends on whether you need the analysis alone or the review around it, and on how much of the work you want to do in code.