Choosing Keywords for a Paper: What They Do, How to Pick.

To choose keywords for a research paper, start from the two jobs they do: they feed the indexing that lets a searcher find your paper later, and on many submission systems they feed the classification an editor uses to match your manuscript to reviewers. Pick terms your title does not already contain, line them up with your field's controlled vocabulary where one exists, and keep the list inside the count your target journal sets. Then test the list by running it as an actual search and seeing whether papers like yours come back.

What do keywords actually do on a submission?

Two things. They add search routes to your paper inside the indexing databases, and on many journal platforms they help match your manuscript to the right reviewer.

The first job is the one most authors picture. A database indexes your title, your abstract, and your keywords, and a keyword adds a way to reach the paper that the title and abstract might miss. The second job is quieter and easy to overlook. Submission systems such as Editorial Manager collect keywords and classifications that describe both the manuscript and the expertise of registered reviewers, then let an editor search for reviewers whose listed subjects match the manuscript's. Taylor and Francis tells its editors to find peer reviewers through predefined keywords the same way. A term that is too vague, or that names your subfield in nonstandard wording, can steer your paper toward a weak reviewer match or none at all. So the keyword field is not only about future readers. It is read by a human editor the week you submit.

How many keywords should a journal article have?

The journal sets the number, and it is usually somewhere between three and eight. Read the author guidelines for the exact range before you draft the list, because the submission form can enforce it.

There is no shared convention across publishing, only each journal's own rule. Frontiers in Neurology, to take one published example, states it plainly in its author guidelines: "Pick 5-8 keywords using a mix of generic and more specific terms on the article subject(s)." Two instructions hide in that one line. One is the count. The other is the spread: some terms wide, some narrow. A study of a rare condition might pair a broad term for the field with the specific syndrome name and the method used. The wide terms catch readers browsing the area; the narrow ones find the few people working on your exact problem. Treat five to eight as this journal's number, not a universal one, and check your own target's range.

Should keywords repeat words from your title?

Generally no. The database already indexes every word in your title, so a keyword that copies a title word spends a slot without adding a new path to your paper.

Think of each keyword as a separate pointer. Your title words are already pointers. Copy one into the keyword list and the slot does nothing. Spend it instead on something the title had to leave out. That is often the method (a "cohort study" or a "randomized controlled trial"), the population you studied, a broader field label, or a synonym a searcher would plausibly type instead of your phrasing. A title that names a disease and a drug still says nothing about the study design, which is exactly why the design earns a keyword slot. The goal for every entry is a route the title did not already provide.

Where do you find the right terms to use?

Use your field's controlled vocabulary. A controlled vocabulary is a fixed, curated list of approved subject terms, and for biomedical work the main one is MeSH.

MeSH, short for Medical Subject Headings, is the thesaurus the United States National Library of Medicine maintains and uses to index MEDLINE and PubMed. Indexers there read the full text and attach roughly ten to fifteen MeSH terms to each article, marking a few as the main topics. You do not choose those, and they sit apart from the keywords you submit with your manuscript. Knowing the controlled term still helps, because it tells you the standard wording searchers actually use, which you can then echo in the free keywords you do control. The MeSH Browser is open to anyone through the NLM site, so you can look up the approved term for your subject before you write the list. Fields outside biomedicine often publish their own scheme; when yours has one, borrowing its exact phrasing makes your paper easier to find than an invented label would. Reading a journal's aims and scope also shows the vocabulary its readers expect.

How do you test whether your keywords work?

Run each keyword, and a couple of combinations, as a search in the database your field uses. Then look at what comes back.

This is the step most authors skip, and it is the one that turns a guess into a checked list. Type a candidate keyword into PubMed, Scopus, or whichever index your community searches, and read the first page of results. A term that returns tens of thousands of unrelated papers is too broad to help anyone find you. A term that returns nothing is either too obscure or phrased in a way no one else uses. The wording you want sits in the middle: it surfaces work close to yours, the neighbors your paper belongs beside. Adjust toward that wording. You are querying the same index your future readers will, so their results are a preview of whether they will ever reach your paper.

A tested keyword list is one small piece of the first screen a journal runs, alongside the abstract length, the reference style, and the count in the keyword field itself. Working through a submission checklist catches these together, and it pairs well with choosing the journal in the first place. A last read against the target journal's own format rules, whether you do it by hand or run the manuscript through Waraq, is where a mismatched keyword count or an off-scope term gets caught before an editor sees it. Fix the list there, and instead of tripping that first screen your keywords earn their slots: they bring the right readers and the right reviewer to your paper.