I use natbib with the following options:


Using the \citep command, I get citation call-outs that look as follows:

(Author et al. 2013, Author & Other 2011, Author et al. 2012)

What I would like is for the list to be compressed and sorted in ascending year order, at least for the same author list (or abbreviation):

(Author & Other 2011, Author et al. 2012,2013)

If I use natbib's sort&compress instead of sort, then I get, at another place

(Author et al. 2015a, 2017, 2015b)

What I would like is, again, that the years be in ascending order:

(Author et al. 2015a,b, 2017)

I think the problem is that the sorting is done according to the order in the references, where all author names matter, including the hidden ones in the "et al." parts.

With author-year citations and abbreviated author lists, this does not make sense, IMO, and is against many journal rules.

Is there a way in which that can be fixed?

  • 1
    Your write-up is somewhat confusing. For example, what's "hiding" behind "(Author et al. 2013, Author & Other 2011, Author et al. 2012)"? If the three underlying entries have author fields "Author and Buthor and Cuthor", "Author and Other", and "Author and Yther and Zther", resp., then "(Author et al. 2013, Author & Other 2011, Author et al. 2012)" would indeed be entirely appropriate and customary. Sorting the items in citation call-outs purely by first author (ignoring all other authors' names) and then by year would be quite unusual, to say the least.
    – Mico
    Commented Mar 31, 2017 at 11:49
  • 1
    What you think is unusual is actually the rule for many journals. I looked up a random example in Global Change Biology (DOI:10.1111/j.1365-2486.2005.00902.x ), and for example "Saleska et al., 1999, 2002" refers to Saleska SR, Harte J, Torn MS (1999) and Saleska SR, Shaw MB, Fischer ML (2002)... So the two Saleska refs are collapsed despite other authors, and this is what I would like natbib to do
    – user52366
    Commented Mar 31, 2017 at 13:52
  • 1
    natbib's behaviour is actually more strange, IMO: it collapses the refs if these follow each other in the references, despite different authors following the first one. natbib seems to apply the logic of numbered refs to author-year citations. Obviously, one can write [1-3,7] only if 1 to 3 are in sequence, and this group will be pulled apart if another reference moves in between (it might become [1-2,4,8]). However, this logic does not apply with the author-year format.
    – user52366
    Commented Mar 31, 2017 at 13:59
  • I suspect that achieving your sorting objective using the natbib package would amount to a new-feature request. (Requests for new features are off-topic for this site -- they're supposed to be directed to the packages' authors and maintainers.) However, achieving your objective may be quite feasible if you can switch to biblatex.
    – Mico
    Commented May 5, 2017 at 4:26

1 Answer 1


This is the standard behaviour of biblatex's authoryear-comp (or authoryear-icomp) style. The sort scheme with that style is nyt, i.e. name > year > title. Citation call-outs are also sorted.

biblatex tries very hard to avoid collapsing different lists of authors into the same abbreviated form. If this is wanted and forced with uniquelist=false, you get exactly what you expect.


  author = {Author, A. and Buthor, B. and Cuthor, C.},
  year = {2001},
  title = {Alpha},
  author = {Author, A. and Cuthor, C. and Buthor, B.},
  year = {2002},
  title = {Beta},




Author et al. 2001, 2002; Sigfridsson et al. 1998; Worman 2002

in citations.

With uniquename=true (the default), you get

Author, Buthor et al. 2001; Author, Cuthor et al. 2002; Sigfridsson et al. 1998; Worman 2002

For help on how to switch to biblatex see What to do to switch to biblatex?, biblatex in a nutshell (for beginners), bibtex vs. biber and biblatex vs. natbib. Since it is recommended to use the new Biber backend instead of BibTeX with biblatex, have a look at Biblatex with Biber: Configuring my editor to avoid undefined citations.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .