Why Zotero's output can still fail the journal's own rule.
Waraq · July 25, 2026 · RULE OF THE WEEK
When Zotero or EndNote formats a bibliography that does not match a journal's requirements, the cause is almost always the style file you selected, not the references themselves. A reference manager applies one output style to the whole document, and that style encodes a fixed set of choices: which base style, how many authors it prints before et al., and which item types it renders. If the style was built for a different journal, every entry comes out tidy and every entry can still be wrong. The repair is to match the style to your target journal's reference rule, then read the exported list against that rule by hand before you submit.
Why does the official journal style file still produce violations?
Because a citation style encodes formatting, not the journal's full submission policy, and it can be the wrong variant or simply out of date. Zotero and Mendeley read styles written in CSL, the Citation Style Language, and those styles live in a shared public repository that anyone can copy from. A journal like PLOS ONE has an official entry there. That does not make it a guarantee.
Two gaps open up. A style file in the repository can lag the current guidelines, so a rule the journal tightened last year has not reached the version you downloaded. And a style only controls how an entry looks. PLOS ONE states its base rule plainly: "PLOS uses "Vancouver" style, as outlined in the ICMJE sample references, with references numbered in order of appearance and cited in square brackets." A style file can reproduce that shape and still let through content the journal forbids, because the ban lives in the guide for authors and not in the formatting. So the output looks correct and the submission fails a check the file was never able to run.
Failure one: the style variant does not match the journal's reference rule
You picked a generic Vancouver style when the journal wants its own. Vancouver is a family of house styles, not a single standard, so two files both labelled Vancouver can number, bracket, and punctuate differently.
PLOS ONE cites references "numbered in order of appearance and cited in square brackets." A generic Vancouver style in the Zotero repository might render citations as superscripts instead, or order the list alphabetically rather than by first appearance. Both look like Vancouver at a glance. Only one matches the sentence the journal wrote down. This is the same trap that catches people choosing between numbered and author-date referencing in the first place: the label tells you the family, the journal's spec tells you the exact member of it. Select the style named for your target journal where one exists, and treat any style with a generic name as a guess until you have checked its output against the guidelines.
Failure two: the et al cutoff is set once and the journal disagrees
The style file carries a single author cutoff for the entire document, and there is no reason it matches the number your journal names. PLOS ONE is specific: "Use the numbered citation (citation-sequence) method and list the first six authors, et al."
In CSL, that truncation is two attributes, et-al-min and et-al-use-first. A style set to list the first three authors before et al. applies three to every reference in your paper, on every journal you point it at, until you change it. Point that same file at PLOS ONE and each multi-author entry loses names three and beyond, which is wrong here even though every printed name is accurate. The reverse happens too: a style that spells out every author sails past the six-author cap PLOS ONE sets. Because the cutoff is one global setting rather than a per-journal one, the same manuscript formats correctly for the journal the style was built for and wrongly for the next one you submit to. Nothing warns you. The et al. simply lands in the wrong place, quietly, across the whole list.
Failure three: preprints and theses slip into the reference list
A reference manager formats whatever sits in your library, so unpublished items pass straight into the bibliography looking exactly as polished as everything else. PLOS ONE draws a hard line: "Do not cite unavailable and unpublished work, personal communications, or rely upon retracted research in the reference list."
This one no style file can catch, because it is a content rule, not a formatting rule. Zotero holds item types for preprints, theses, manuscripts, and personal communications, and the style renders each of them in the correct numbered shape. The output is flawless and the reference is still forbidden. A preprint you cited in an early draft, a thesis from a collaborator, an email exchange logged as a personal communication: each one formats cleanly and each one trips the rule the moment an editor reads the list. This is where a manuscript that looks finished draws the desk reject you could have seen coming, because the problem was never visible in the formatting.
How do you check references against the journal's own rules after export?
Read each entry against the journal's reference section after export and before submission, one rule at a time, treating the CSL file as a draft rather than a verdict. Three passes cover most of the damage. First, does the citation shape match the variant the journal names, brackets and numbering order and all. Second, does the author cutoff match the number in the spec, not the number your style happens to carry. Third, does anything unpublished, retracted, or personal sit in the list where the journal bans it.
That last pass is the one no export can run for you, and it is the one editors notice. Checking the finished list against the journal's written rules, not against the style file that produced it, is the pass Waraq runs on a manuscript before it reaches a reviewer. The official style file gets you most of the way. The rule is what you are judged against, so read the list against the rule.