Heading levels: PLOS ONE allows three, Frontiers allows five.

There is no single heading-level limit that holds across all journals; the number is set per journal and lives in the guide for authors. PLOS ONE caps a manuscript at three heading levels. Frontiers in Neurology lets you go to five. Before you commit to a four-level outline, check the one journal you are submitting to: a depth that passes at one publisher trips the template at another.

Do journals actually write down a heading-level limit?

Yes, and the two we can quote sit far apart. A heading level is one step of nesting: the section title is level one, a subsection under it is level two, a sub-subsection under that is level three, and so on down.

PLOS ONE states the rule plainly in its submission guidelines: "Limit manuscript sections and sub-sections to 3 heading levels." Frontiers in Neurology, in its author guidelines, allows more room: "You may insert up to 5 heading levels into your manuscript (i.e. 3.2.2.1.2 Heading Title)." That is the same structural decision, made twice, landing two levels apart. Neither number is a suggestion. Both are stated limits the editorial office can check against your file, the same way it checks an abstract that runs over the word count.

Why do extra heading levels break the journal template?

Because a production team styles each level with a fixed set of typeset formats, and a level past that set has nowhere to land. When PLOS ONE says three, the house template carries exactly three looks: a large section head, a smaller subsection head, a run-in sub-subsection head. That is the whole vocabulary.

Add a fourth level and the typesetter faces a choice with no good answer. They can invent a format the journal never approved, or they can flatten your level four by hand into level three and hope the meaning survives. Both cost copy-editing time, and both happen after acceptance, when you have the least leverage to argue about how your structure reads. Frontiers can absorb a deeper file because its template defines five looks.

How do you flatten a four-level outline without losing the argument?

Three moves cover almost every case: merge the sibling branches, promote the one that carries weight, or demote the rest into lead-in sentences. Pick per branch, not once for the whole document.

Merge first. If your level-four headings are three short sub-subsections that each run a paragraph, they are usually one idea split too fine. Fold them into a single level-three subsection and let the paragraph breaks do the separating. Promote second. When one level-four heading is doing real work and the others around it are thin, pull that one up to level three and absorb its weaker siblings into the running text above it. Demote last. A heading that only labels a paragraph can become the first few words of that paragraph instead, set in bold, so the signpost stays and the level disappears. That bold lead-in is body text as far as the template is concerned, which is exactly why it does not count against the limit.

None of the three loses an argument. They lose a layer of scaffolding the reader was never reading as a layer.

Do the Abstract and References count as heading levels?

No. The Abstract and References are top-level manuscript sections; each sits at level one. The heading-level limit governs how deep you nest beneath a section, not how many sections your manuscript has.

PLOS ONE spells out the section order separately: title page, Abstract, and Introduction open the manuscript; Materials and Methods, Results, Discussion, and Conclusions form the middle; Acknowledgments, References, and supporting-information captions close it. Every item on that list is a level-one head. You can have a dozen of them and still be inside a three-level limit, as long as nothing nested underneath reaches a fourth tier. The count that matters is the depth of your deepest branch, measured from its section head down.

How do you check heading depth before you upload?

Open your outline view and count the deepest branch, then compare that number to the one in your target journal's guide for authors. If the deepest point is level four and the journal says three, you have a fix to make before submission, not after.

Most word processors will show you this in seconds. In the navigation or outline pane, the indentation depth is the level depth: the furthest-indented heading is your true maximum, and it is often deeper than you remember making it. Watch the branches you revised late, since a heading added during revision is where an extra level usually sneaks in. A structural slip like this rarely stops a manuscript on its own, but it travels in the company of the format problems that get papers desk rejected, and an editor who spots one tends to look harder for the next. When Waraq checks a manuscript against a target journal's rule set, a fourth level in a three-level journal gets flagged with that journal's own wording quoted in the margin, so the note reads "Limit manuscript sections and sub-sections to 3 heading levels" rather than a vague comment about structure.

Do the count before you polish anything else. A four-level outline built for a five-level journal reads fine until you send it to a three-level one, and by then the reorganizing is harder than it would have been on the whiteboard. The same care applies to the smaller rules that stack up at submission, from heading depth to how many authors you list before et al. Pick the journal, read its number, and match your outline to it while the structure is still easy to move.