What reference manager is the best option?


Managing references can be a major pain in the butt. It’s one of the more annoying parts assembling a manuscript, especially when you have to reformat after a rejection.

So, what’s the most efficient way of managing references for a manuscript?

Some of the options people use are BibTeXEndnote, Mendeley, PapersReference ManagerZotero. Or you could just keep a big list of references in a word processing file.


At one point, many years ago, I used Endnote. But they end up releasing new versions of the software twice as fast as new editions of Campbell’s Biology and it’s just stupid to keep paying a company over and over for the same damn product. A sucky, glitchy, hard to use product, by the way. If you’re not using Endnote, for the love of some deity, don’t let yourself be tempted.

I’ve been looking for a New Way for the last year or so, after year or two of doing it old school. I’m writing in Word to share files with collaborators. And I’m not paying money for a program that will go out of date next year. Both Mendeley and Zotero are free. But Mendeley was bought by Elsevier last year, which has a number of downsides. I thought I’d go for Zotero. But I had a difficult time importing my old endnote library (and my endnote version expired and wouldn’t let me do it, long story). In the Mendeley there already are a huge set of collected references, and its integration with web searching for new citations is super-smooth.

So, I went with Mendeley.

Two weeks and one manuscript later, I think it’s coming along pretty well. It plays nicely with my version of Word and crashes infrequently enough that I haven’t yet damaged my laptop by punching it.

As far as I can tell, a lot of folks are using Mendeley. But I’m really curious to see the results of the poll.

Any thoughts about reference managers, what works and what doesn’t, and if you’ve changed from one system to another, any good ideas? Or unsubstantiated opinions?

Nota bene: If you write your manuscripts in LaTeX, I don’t think I can really handle hearing even more about its awesomeness. I take your Word for it.

34 thoughts on “What reference manager is the best option?

  1. There’s two questions here. What do I use? BibTeX. Which is the most efficient? Mendeley. I recommend Mendeley to all our PGs, and actually teach a workshop on it, despite not using it myself. Which begs the obvious question, to which the annoying answer is that if you spend your life in LaTeX then it’s just easier to have a compatible tool. I wouldn’t recommend BibTeX to anyone who wasn’t already planning to make that additional time investment.

  2. Inertia, collaborator preference, and opportunity cost keep me using Endnote rather than switching to mendeley but am open to being convinced to switch.

  3. I’m in the same boat as Emilio. When I get a new computer in the next few months, I may switch to Mendeley rather than being forced to upgrade yet again. I don’t think I’ve used any of the new ‘features’ added by Endnote since v4. All I use it for is Cite-While-you-Write.

  4. I hear Mendeley is really popular. I’ve been using Papers for years (version 2 even though I know Papers 3 is out but have only heard awful things. It does it’s job well enough. If I were to switch, it’d probably be to Mendeley. I would love one that integrates well with Google Docs. I think I heard something about Zotero doing that well. And if you’re an Evernote power user, I wonder if that could be fenagled into a reference manager of some kind.

  5. I went from Endnote to Mendeley years ago. I pay the lowest paid level, which allows me additional space, and have groups for each class that require students to do bibliography searches. I invite students so they can use my references and also add theirs. Mendeley has a great search function, and allows some social functions too (following, joining groups etc).

  6. I use Zotero, and export that to use in Bibtex. I’ve really enjoyed Zotero, but I started my reference library fresh with it years ago, so didn’t have to worry about importing from EndNote.

    I find it’s quite powerful; easy import from the web, really good organization tools, and nicely extensible. Haven’t really tried anything else, though, since a very brief flirtation with EndNote in undergrad.

  7. I use EndNote. I tried Mendeley and Papers and found them slower and buggier than EndNote. EndNote’s Cite-While-You_write is quite fast. I also had a difficult time modifying output styles, italicizing species names etc. This was a few years ago so maybe there has been improvements since that time. I also like that most journals have downloadable styles for EndNote. Finally, EndNote really syncs well with Web of Science. I know a lot of people complain about EndNote but I really haven’t had problems with it – although I agree that the constant $$ for updates is very annoying.

  8. I write 99.99999% of my manuscripts in LaTeX, so I use BibTeX. Actually I use BibTeX with a front-end since I have better things to do than manually edit BibTeX files. :) But I’ve yet to find a BibTeX front-end that is better than tolerable (and often I find myself manually editing BibTeX files anyway). I’d love for someone to come up with a BibTeX program that’s easy to use with a great, intuitive interface.

  9. I switched to Mendeley a few years ago. The only issue I’ve had is that it handles non-journal publications poorly. I cite a lot of reports and always have to hand edit the entries in the bibliography. I did not have the same issues with EndNote.

    I’ve also tried creating my own citation styles in both EndNote and Mendeley (http://escsecblog.com/2013/02/18/formatting-your-references-for-the-canadian-entomologist-using-mendeley/ ) . The process is much easier in EndNote, Mendeley’s Virtual CSL editor is not easy to understand, or use.

  10. I have never had major problems with EndNote, though I know other people have. I like it a lot. I think buying it was one of the best purchases I ever made as an academic.

  11. I use Papers, and really like it. I started using it because there was a free copy available through my MSc. lab, so I tried it out and haven’t looked back. I suspect that’s the case for a lot of people – if you try something and it tends to work, the idea of switching is a big hurdle to overcome, even if you hear great things about another system.

    Interestingly, we took a poll in my “welcome to grad school” type seminar, and >75% of the class didn’t use any manager, and was convinced they wouldn’t ever need to. Would love to see how that changes by the end of the degree!

  12. I voted LaTeX/bibtex. But neither is a reference manager. I simply meant that for me bibtex is the best database format for portability . But using the data may not involve “bibtex” the ancient program at all.

    Re frontends (@acdalal): Jabref is fine for a bibtex file front end, but I use emacs most of the time since I’m in emacs (refetx) any time the computer is on anyway. Do you use jabref and what would you improve if so?

    I still use LaTeX a lot but more and more I write plain text in org-mode or markdown and use pandoc with bibtex.

    I wrote a script to pull all of my refs from endnote and save as bibtex back in 2000 or 2001. Fourteen years later, I have never touched endnote again.

  13. I had to switch from Mendeley to Zotero in the middle of submitting a paper with 150 references, because the more references I added, the slower Mendeley became, at one point it would take several minutes until I had added a reference. It crashed too, but the slowness was what made me want to punch my computer.
    I filed a bug report for this and I’m still tracking it after I left Mendeley for good.
    They don’t seem to have fixed it.

    Zotero didn’t have this problem and I’ve been happily using it since. I use the standalone client, since the plugins tended to slow my browser down if I had a big library (a while ago, maybe they fixed it).
    At least Mendeley didn’t lock me in that badly (though my notes and folders were lost).
    I think it’s unfortunate that neither of those apps has a social component (so if thousands of users fix a buggy reference, they don’t take notice) and neither does stuff like changing title case/sentence case according to CSL. Probably doing that properly is hard, but they could at least give warnings, when one of your 150 refs is in title case.

  14. @dylan, I’ve used JabRef extensively (am using BibDesk now, just because). My big issues with it are: (1) I’ve never been able to successfully automatically pull down references from citeseer and the like, (2) fonts are always weird and attempting to change the fonts never has the desired effects, (3) (perhaps related to the fonts issue) every once in a while weird characters are inserted into entries and then hilarity ensues while I attempt to find them. So, nothing that’s a deal-breaker, but little things that annoy.

  15. I use Endnote mostly from inertia. Ctrl+M, Ctrl+F, Alt-1, Alt-2 just roll of my fingers at a subconscious level. I agree with the annoyance of frequent upgrades and the fact that most of the newish web-integration features are not things I need (indeed the innately web-oriented Mendeley scares me for reasons I can’t quite put my finger on). However, once I shut off CiteWhileYouWrite I find it incredibly stable – meaning crashes/loses edits like once a year or less and never lost references in 10 years (but incredibly unstable for me with CiteWhileYouWrite on) and they do have real tech support so I haven’t had any trouble transferring licenses for old copies to new computers.

  16. I use Papers 3.x to read and organize PDFs but write in LaTeX so create a Papers directory for each manuscript which includes all the citations in that ms and then export these to bibtex. Papers 3.x would be fantastic if it didn’t crash everyday. And if it searched for authors. And if the ipad Papers app synched faster than 3 days (yes you read that correctly). But I have about 6K PDFs and need some way to digitally organize these and Papers does a good job with that so I stick with it.

  17. A lot of universities will provide a program (usually Endnote) to faculty/staff/students for free. Unless you are tenured and are sure that you want to stay forever, don’t be tempted by any such nonsense. Eventually you will leave and be faced with the choice of re-entering your references into a new program (bad) or paying to continue to use the program you already have (worse). For anyone just starting their library now, just use open source software! This goes for other software, too; don’t invest the time to learn your adviser’s favorite stats program (or whatever program your university supplies) unless it’s something that you’ll have access to after graduation! There are plenty of powerful open source options. I like Zotero for references and R for stats.

  18. zotero all the way here. free, fast, portable, and lets you install about any citation style you could want. I don’t see the downside.

  19. I use Mendeley, but I have some gripes:

    1) It imports absolute shit for most JSTOR papers (Ecology, AmNat, etc.) and a few other services (and all scans) – unless it is recent, you pretty much have to enter everything yourself – even when it can find the reference with the title search function, it usually deletes authors or puts erroneous volume numbers and years.

    2) As mentioned above, it struggles with books/chapters/non-journal references. It won’t even let you put in all the necessary information sometimes.

    3) This is a personal gripe that probably everyone doesn’t have, but I put all my papers in one references folder, which Mendeley grabs from. This is not the most efficient way to do it – as I don’t actually cite most papers I read (seminars/classes/general interest) – yet in order to find the right ones for citing, I need to go through all the erroneously labelled ones and fix them – this adds hours each time I do a bibliography (if I published papers more often, this would be less of a problem…).

  20. Some folks convinced me to try zotero and I love it. I just submitted a paper where me and a co-author were working on the same draft in google drive and had a shared library. It was the most seamless writing I’ve ever done. I’m also using for a big collaborative review/meta-analyses and the library sharing is great. I just need to figure out how to import my endnote library to make the full switch.

  21. Hearing that Papers and Mendeley crash more than EndNote is discouraging. I use CiteWhileYouWrite in EndNote but with the automatic formatting turned off, which seems to result in fewer crashes.

  22. I use CiteULike (www.citeulike.org). I switched to Mendeley briefly and promptly switched back. I prefer CUL mostly because it is HUGELY faster than Mendeley – it has an HTML-based interface instead of Mendeley’s Javascript-based one. Which means your browser can cache pages and you don’t have to wait for minutes while a page displays. With Mendeley I would sometimes follow a link, hit “back,” and then sit growling in frustration while the page slowly rebuilt.

    Also CUL makes it super easy to point people to your list of references — something Mendeley STILL does not do (WHY?!).

    Biggest downside to CUL is the lack of integration with word processors for adding citations to papers. To do that I have to export to Zotero and use them. This is annoying, but it’s worth it to me (I’ve also tried Zotero but I still really prefer CUL’s fast entirely web-based interface).

    Honestly if one of these apps had a decent Android app I’d probably switch to them, but none of them seem to. Perplexing. So I end up using CUL’s web interface on my Android — works OK but they don’t have good small-screen support so it’s annoying. Why isn’t there better mobile support for these apps? Doesn’t everyone else read PDFs on their mobile like me these days instead of on a computer?


  23. since i started my PhD studies back in 2008 i’ve always been using zotero. although there are minor problems from time to time my general feeling is that it’s running very well. i especially like its compatibility with MS word AND open/libreoffice

  24. Within 24 hours of my comment above, Papers 3.x lost a top level Collection, one that had many many sub-collections (and sub-sub, etc.). The pdfs aren’t lost just the way to navigate to it via the collection mechanism. I put a huge amount of effort organizing my pdfs this way because it is was the way I chose to re-call certain key papers. Again, I mostly need a way to organize and read pdfs that synchronizes between my desktop, macbook, ipad, and iphone (I haven’t printed a pdf in over a decade). I abandoned Word a year ago (highly recommended) and export collections to bibtex, so don’t need a functional cite-while-u-write feature.

    I have spent the last 3 days exploring Mendeley and Zotero. I thought Mendeley was quite good but I have concerns over its future. I am impressed with Zotero and like the open-source ethic. I have been playing with a sample set of pdfs, organizing them with one level of collections but mostly with tags. Tags work elegantly as long as you take the time to tag a paper. Unfortunately, the ipad reader Papership does not support tags well.

  25. As I mentioned on twitter when Terry initially tweeted about this topic, I have moved to using Paperpile, an online system that one access through the Chrome web browser. I have a local copy of my PDFs but these are archived on Google Drive so that I have access to them anywhere I have access to Chrome.

    I’ve been very happy with this although I’ve only been using it for 6-8 months. Before that I used to just have folders of PDFs and one large, plus a few smaller, BibTeX files. Paperpile integrates so nicely with web searches and publishers’ websites that I do wonder how I managed without it before hand.

    Paperpile is a subscription-based model, so I do have to stump up $ annually to keep using it, but thus far it has been well worth that investment.

  26. OK, I’m going to put my hand up and say that I’m one of those who doesn’t use a reference manager. Perhaps it’s a generational thing: they didn’t exist when I was researching my PhD in the early 90s, or if they did I wasn’t aware of them.

    But there are plenty of scientists older than I who use one, so age is not the whole story. I did try using Endnote about 15 years ago (if I recall correctly) and got disillusioned when I lost the whole database when a laptop I had broke down. I’d taken quite a while to input a lot of older literature and couldn’t face doing it again.

    I’m aware of the arguments around saving time formatting references for different journals, but in my experience as an editor and reviewer (and monitor of my personal citations on Web of Knowledge), it seems to me that the time saving comes with a trade-off in accuracy. I have the strong impression that the referencing error rate has increased over time, and I think it’s due to the use of reference management systems. Scientists often do not take enough time over checking the accuracy of the references they use. I regularly have to correct inaccurate citations in WoK, and see this all the time in manuscripts.

    It’s not just a matter of wrong page or volume numbers: I also see mis-citation of my work, e.g. making reference to a paper that does not say what the authors claim it says, and it’s clearly been mixed up with another of my papers which does.

    So what, you might say? The odd wrong reference is hardly a big deal. Maybe so, but attention to detail and accuracy is such an important part of any scientist’s training. If basic errors are creeping into something s simple as a reference list, what does this suggest about the quality of data analysis, accuracy of figures and tables, etc? Is reliance on technology coming at the cost of attention to detail and accuracy?

    As I said, this is just an impression and I know of no study that has assessed whether there has been an increase in the rate of mis-citation over time. Certainly the literature in the pre-computer age was not immune to mistakes.

    I’d be interested to know what others’ views are on this.

  27. I use mendeley. What I like is that it is free, and that it automatically syncs (in my case to dropbox) papers I save with intelligible names so that I can access them from anywhere, also with a tablet. What I dislike is that the journal formats are very bad and that I can’t figure out how to change them. I used to use endnote and hated it, definitely not worth spending any money on now that there are better alternatives.

  28. I’m responding to Jeff’s remark/question. With 0% data and 100% notions and hunches.

    One of the big arguments for Endnote, back in the day when it was emerging, was that it would actually correct the mis-citation errors that you would find in the literature. In theory, people would re-write out their citations from their copies of the papers and get all of the correct information in their manuscripts. But in practice, people would copy citation information from other papers rather than look at the original document (it was easier, after all.). So citation errors sometimes magnified, if a key paper had a mis-citation in its refs cited section, that mis-citation would appear in others. I noticed this on occasion, when I was hand-typing my references from papers and notice they didn’t match when I was reading other ref cited sections, some page number would be off, or the title was slightly wrong. (I don’t see that much any more, but that doesn’t mean much because I wouldn’t have the opportunity to notice that I do it differently.)

    Now with reference management systems – and those with online sharing like Mendeley – we are using the common reference information but it can be modified to be corrected. But most often (at least for me), the citation information is exported from the website of the journal itself, where the information gets imported into the reference manager. (This is actually way faster and more convenient than old-schooling it. I’ve spent more of my career doing the old school and am glad I’m not doing it anymore.)

    Now that you can find the correct citation info for a paper using its name and the internets, I’m not so concerned that citation information might be off. Earlier, a wrong volume number or page number could result in real difficulty in finding a paper. I’ve spent lots of time plying the stacks of the library in Boulder trying to find a couple papers with bad citations (so that I could photocopy it. It wasn’t that long ago, folks.) So, the cost of these errors is lower now, but also, I think the probability is lower as long as the publishers have the right information in their databases.

  29. Hi Terry,

    We’ve two contrasting hypotheses here – either the error rate in reference lists is improving or getting worse, with a null that there’s been no change. I hope that at some point someone works out a way to test this without having to do it by hand! Until that day we won’t know which is correct.

    Your final point however, about being unconcerned about wrong citations because it’s easier now to track down a reference, misses part of the point. If we agree that citation rates of papers and corresponding h-indices are important metrics for us as scientists (a debatable point, I accept) then mis-cited references represent a loss of academic impact because they are not credited to us on databases such as Web of Science. This can be particularly significant for early career researchers where a few lost citations can make a difference of several points on their h-index.

    These errors can be corrected of course by using the appropriate form on the databases, but many researchers don’t bother or are unaware.


  30. Concerning the ability to add papers while searching the web, the simplicity in organizing them, to work at the same paper with your lab and the easy way to download them to my library i tried loads of different reference managers. In the end I only was convinced by paperpile.com which won them all.

Leave a Reply