Writing a Title That Gets Your Paper Read

Every reader and every database meets your paper through its title. It has to say what you studied, in whom, and how, without claiming more than the study shows.

Guide · 10 min read · Directive Publications Editorial Team

What a title has to do

Every reader of your paper meets the title first, and so does every database that indexes it. The title has to do its work before anyone reaches your abstract.

The ICMJE Recommendations describe the job precisely: the title "provides a distilled description of the complete article and should include information that, along with the abstract, will make electronic retrieval of the article sensitive and specific."

Sensitive means that someone searching for your topic finds your paper. Specific means it does not turn up in searches it cannot answer, where it only wastes a reader's click.

In PubMed, a search tagged [ti] matches only words in the title, and [tiab] matches words in the title, abstract and author keywords. PubMed's help pages also warn that filters relying on MEDLINE indexing, such as article type filters, may exclude citations that have not yet completed indexing. A text search can match only the words you actually chose.

Human readers need the same thing, whether they are editors or clinicians scanning search results. The test: could someone who reads only your title decide correctly whether your paper is relevant to them? If they would have to open the abstract to learn the population or design, the title is not finished.

Build the title from its parts

Most medical research titles are assembled from the same few pieces:

• Topic: the intervention, exposure, test or condition at the centre of the study
• Outcome: what you measured
• Population: who was studied, plus the setting only if it matters
• Design: how you studied it
• Comparator: only when a reader cannot infer it

The AMA Manual of Style team reduces the core to a formula on its AMA Style Insider blog: the key variable or variables measured, plus the individuals or condition studied.

Here is how that works on a hypothetical study, invented for illustration: a cluster randomised trial of pharmacist-led medication review for older adults, with unplanned hospital admission as the outcome.

1. Medication review in elderly patients. Topic only: no outcome, no design, and vague about who was studied.
2. Pharmacist-led medication review and unplanned hospital admissions in older adults. Intervention, outcome and population are now all searchable.
3. Pharmacist-led medication review and unplanned hospital admissions in older adults: a cluster randomised trial. The design tells a reader what kind of evidence to expect.

Notice what never made it in: "A study of", the usual-care comparator, the region and the recruitment years. All of that belongs in the abstract.

Name the study design where guidelines expect it

The ICMJE Recommendations note that reporting guidelines recommend, and some journals require, design information in the title, "particularly important for randomized trials and systematic reviews and meta-analyses." The main guidelines differ in how firmly they ask:

Where a guideline allows either, house style decides, and placement varies. PLOS ONE asks for the design in the subtitle for clinical trials, systematic reviews and meta-analyses. AMA style guidance accepts subtitles such as "A Randomized Clinical Trial" for trials and reviews, but says observational studies should generally avoid subtitles.

Use the design term people search for, such as "cohort study" or "systematic review and meta-analysis", in the journal's spelling of randomised or randomized. And check the current checklist version before you cite it: CONSORT 2025 supersedes CONSORT 2010, and the EQUATOR Network library lists current guidelines by study type.

Descriptive, declarative or question titles

Here is the same hypothetical trial titled three ways. The "result" in the declarative version is invented for this example.

"Informative" is simply another name for declarative. Wager, Altman, Simera and Toma define declarative (or informative) titles as those that "state the study findings", and descriptive (or indicative) titles as those that describe the study topic and design.

Journals disagree. Wager and colleagues note that many journals prohibit declarative titles while a few encourage or even require them. AMA style guidance advises against revealing results in the title, including declarative statements such as "X is associated with Y".

The evidence is thin. In a randomised trial by Wager's group, students and doctors read a fictitious trial abstract under either a declarative or a descriptive title, and the authors found no evidence that the declarative title changed readers' confidence in the conclusions. They tested abstracts, not full papers, and limit their reassurance to reports of randomised trials. Citation studies are observational: Paiva and colleagues (Clinics, 2012) linked short titles presenting results to more citations, and Jamali and Nikzad (Scientometrics, 2011) found PLOS articles with question titles tended to be downloaded more but cited less.

Use these decision rules:

1. Follow the journal first. Read the instructions, then look at research titles in a few recent issues. If every one is descriptive, expect a declarative title to be changed.
2. Use a declarative title only when the design can carry the claim. The primary outcome of a randomised trial may; a subgroup or pilot result rarely can.
3. Keep causal verbs out of observational titles. "Reduces", "prevents" and "improves" claim more than a cohort or cross-sectional design can show.
4. Use a question only if the journal accepts it and the paper answers it.

Length: say it once, then stop

There is no universal limit. The ICMJE Recommendations set a length only for the short title, and note that submission systems may restrict title characters. JAMA Network editors aim for research article main titles of no more than 100 characters with spaces. Other journals count differently, so check whether the limit is in characters or words.

The research leans towards shorter titles, with caveats. Letchford, Moat and Preis (Royal Society Open Science, 2015) found that journals publishing papers with shorter titles received more citations per paper, but could not establish cause. One possibility they raise is that high-impact journals restrict title length.

Cut words that carry no information:

• "A study of", "An investigation into", "Report of"
• "Effect of" or "Association of" as opening words, which AMA style guidance suggests skipping
• A comparator the reader can infer
• Location and study years, unless they are the point of the study
• "Novel", "new" and "first"

A worked cut, again on an illustrative title:

The main title before the colon is 81 characters and still says what was studied. That matters: AMA style guidance notes that, depending on a journal's digital practices, only the main title may appear in a Google search result. Cover the subtitle; if the rest does not say what you studied, rewrite it.

Words and habits that cost you readers

If a reviewer says your conclusions go beyond your data, check the title as well as the discussion. An overstated title repeats the problem in every database record and reference list.

The short title, or running head

Some journals ask for a second, shorter title on the title page or in the submission system. It may be called a short title, running title or running head, and some journals print it at the top of the article's pages. The ICMJE Recommendations say it is usually no more than 40 characters, including letters and spaces, but journals set their own limits. It never replaces the full title, which is what appears in the citation.

To write one, keep the topic and outcome and drop the design, population and qualifiers. For the hypothetical trial above, Pharmacist review and admissions is 32 characters. Avoid squeezing a term into an abbreviation to fit unless the journal accepts it.

A title check before you submit

Write a working title early, but settle the final one after the abstract and conclusions are finished, because the title has to match what the paper concludes. Then check:

1. Topic, outcome and population named in plain words
2. Study design named where the reporting guideline and journal expect it
3. Style matches research titles in recent issues of the target journal
4. No causal verb unless the design supports the claim
5. Within the journal's length limit, counted the way the journal counts
6. Main title makes sense with the subtitle covered
7. Short title within its limit
8. A PubMed [tiab] search on your key terms shows the words related papers use, and your title uses them too
9. Identical title on the title page, manuscript, submission form and cover letter

That last point slips easily after several rounds of edits; our guide to what editors look for in a cover letter recommends giving the full title exactly as it appears on your title page. For the rest of the package, work through our manuscript submission checklist.

If you spend extra time anywhere, spend it on the first test: read only your title and ask whether the right reader would know to open the paper. The Directive Publications For Authors page links to its author guidelines and article type definitions, so you can check how to prepare your manuscript before you submit.