This question gets more attention than it deserves, because the products are closer to each other than the arguments about them suggest and because the cost of being the only person in your group on a different one is paid every time you share anything. This page is about the four things that actually differ, and about the much more consequential question of what your reference manager does not do at all, which is where a review most often comes unstuck.
Best reference manager: the ranking, and the criterion
- Zotero: Free and open source, strong web capture, and the one most likely to be supported by a university library. The default answer unless your group has already chosen something else.
- Mendeley: Free with paid storage tiers, a competent PDF reader, and a large installed base, which matters mainly because it makes shared libraries with co-authors easy.
- EndNote: Paid, long established, and still the institutional standard in several disciplines. Worth choosing when your department supports it and worth avoiding when it does not.
- Paperpile: Paid and built around a browser and Google Docs workflow. Suits people who write in Docs and dislike desktop plugins; less suitable if your co-authors are all in Word.
Compatibility beats features
Your supervisor, your co-authors and your library between them have already made this decision, and matching it removes a category of friction that no feature makes up for: shared libraries that work, a colleague who can fix your citation style in two minutes, and institutional support when something breaks. Start there, and only look at the four differences below if you are genuinely free to choose.
The four things that actually differ
How well it captures a reference from a page without hand-fixing it afterwards. Whether syncing across devices is included or priced. How gracefully the word-processor plugin handles a document with several hundred citations, which is where the cheap ones show their age. And what happens to a shared library when the person who created it leaves the institution, which matters more than anybody expects.
What none of them hold
Not one of them has a field for why you excluded a paper. They are built around a citation, not a decision, so the record a systematic review is judged on lives somewhere else by construction. That is not a criticism of them; it is the boundary between two layers, and knowing where it falls stops you looking for a screening feature that is never going to appear.
Questions people ask about best reference manager
Which reference manager should I use?
The one your co-authors and library already use, unless you have a specific reason not to. The differences between the main products are small next to the cost of being the only person on a different one.
Can a reference manager run my screening?
No. None of them holds a reason for an exclusion, because their object is a citation rather than a decision. That gap is the whole reason screening tools exist as a separate category.
Should I de-duplicate before or after screening?
Before, always. A duplicate that reaches screening is read twice and sometimes decided differently, which is a disagreement you then have to resolve for no reason.