Your keyword list is a handful of extra routes to your paper. Spending them on words already in your title wastes them, because databases search the title anyway.
What your keywords do after you submit
Keywords are often the last thing entered in a submission form, and the quickest way to fill the field is to lift phrases from the title. That wastes it, because your keyword list does three separate jobs and repeating the title does little for any of them.
It adds search routes in databases. When a publisher supplies author keywords to PubMed, they appear under the abstract and become searchable. The PubMed User Guide says the author keyword field is searchable with the Title/Abstract [tiab], Text Word [tw] and Other Term [ot] tags. So when someone searches title and abstract words, your keywords sit in the same pool as your title. A keyword that repeats the title adds nothing to that pool. A keyword that adds a synonym your title and abstract never use gives the paper a new way to be found.
It helps route the manuscript. Submission systems can use subject information to suggest editors and reviewers. Editorial Manager's help documentation, for example, says that classifications and keywords "enable the system to help match Editors and Reviewers to new submissions."
It labels the paper for readers. PubMed displays publisher-supplied keywords with the abstract, so a reader checking a record can see at a glance what the paper covers.
What keywords do not do is set your MeSH terms. NLM's FAQ on indexing for MEDLINE says that as of April 2022 all journals indexed for MEDLINE are indexed automatically, with human review and curation as appropriate, and that automated indexing is currently based on the title and abstract. A careful keyword list cannot rescue a vague abstract.
Put your core terms in the title and abstract, and spend your keywords on the synonyms, outcomes and settings the title leaves out.
How PubMed and Google Scholar treat your words
PubMed and MEDLINE
MeSH is NLM's controlled, hierarchical vocabulary for indexing and searching biomedical literature, and it is updated every year. Each heading carries a list of entry terms, which are the synonyms that point to it. That list matters when someone searches. PubMed may add MeSH terms, British and American spellings, singular and plural forms, and other synonyms to a query, mainly through its Automatic Term Mapping. "Heart Attack" is an entry term for the MeSH heading Myocardial Infarction, so a search for heart attack can also retrieve papers indexed under Myocardial Infarction.
Not every PubMed record gets MeSH terms, though. The PubMed User Guide notes that some citations are not indexed for MEDLINE, such as preprints, citations published online ahead of print, and articles that reach PubMed from journals not indexed for MEDLINE. For those records, subject searches depend on the words in the title, abstract and keywords. Look up your target journal in the NLM Catalog before you submit: its record states the current indexing status, for example "Currently indexed for MEDLINE."
Either way, NLM's advice to authors is direct: if you want your article retrieved by your preferred terminology, make sure those words appear in the title or abstract, where they are searchable as text words.
Google Scholar
Google Scholar's About page says it "aims to rank documents the way researchers do, weighing the full text of each document, where it was published, who it was written by, as well as how often and how recently it has been cited in other scholarly literature." The keyword field is not mentioned. You cannot influence citations on submission day. You can make sure the terms people search for appear in the text Google Scholar weighs.
A six-step method for choosing keywords
1. Read the journal's keyword instructions
Open the target journal's instructions for authors and note four things: how many keywords it allows, whether it wants MeSH headings or free text, the format it expects (separators, capitalisation, spelling), and whether the submission system also asks you to pick subject classifications from a fixed list. Requirements vary between journals, so a list written for one journal often needs trimming or reformatting for the next. Keywords belong on the same list as your title page and declarations; our manuscript submission checklist covers the rest of the package.
2. List the concepts your title leaves out
Break your study into concepts: population, intervention or exposure, comparison, outcome, design and setting. Mark the ones your title already names. The unmarked ones are your first candidates. For each concept, add the words a searcher in your field would actually type, such as the clinical term and the everyday term, or the full name and the short name.
3. Look up the MeSH headings
Paste your abstract into MeSH on Demand, NLM's free tool that suggests MeSH terms for a piece of text and lists similar PubMed articles. Treat the output as a shortlist, not an answer. The tool's page says its suggestions are machine-generated with no human review, that it may recommend terms not found in your text, and that its MeSH data is static as of the 2024 production year. It also says the tool is not intended for sensitive or protected health information, so paste the abstract, never anything that could identify a patient.
Then confirm each suggestion in the MeSH Browser. Check that the scope note describes what you actually studied. Use the most specific heading that fits, and copy the term labelled "MeSH Heading" rather than an entry term, as NLM's guidance for authors advises. Read the entry terms as well, because they show you the synonyms other people search with. The similar articles list is worth a look too: open a few and note which keywords their authors chose.
4. Cut repeats, broad terms and unclear abbreviations
Now shorten the list. Remove:
- Words already in your title. In PubMed they add nothing to a title and abstract search. Keep one only if your journal's instructions specifically ask for it.
- Single broad words such as "diabetes", "cancer" or "health". They match so many papers that they rarely help anyone find yours.
- Abbreviations with more than one common meaning. MS can mean multiple sclerosis, mass spectrometry or mitral stenosis. Springer Nature's author tutorial on keywords lists a bare abbreviation among its examples of poor keywords.
5. Test each term in PubMed and Google Scholar
Search each remaining candidate on its own and ask two questions.
- Do the first results look like papers yours belongs beside? If they come from a different field, the term is ambiguous. Make it more specific.
- Does anyone use this term? If it returns very little, your field probably uses a different word. Find it and use that instead.
In PubMed, open Advanced Search and expand the details next to each query in the History table. Search Details show how PubMed translated your term, including any mapping to MeSH, which tells you whether your wording and the controlled vocabulary line up. While you are there, notice which journals publish the closest matches. If they are not the journals you had in mind, that is useful evidence for choosing the right journal as well.
6. Carry essential terms into the abstract
Go through your final list and pick out the terms that name what the paper is fundamentally about, usually the outcome, design and setting. Each of those should also appear in the title or abstract, because that is the text MEDLINE indexing works from and part of the full text Google Scholar weighs. If one is missing, add it where it fits naturally; the methods sentence of the abstract is often the right place for design and setting. Do not stuff the abstract with search terms, though: the first person to read it is an editor deciding whether it goes to review.
A keyword that also appears in the abstract is not wasted, since it still labels the paper and can feed reviewer matching. But when two candidates compete for the last slot, prefer the one your title and abstract do not already contain. Then enter the list in exactly the format the journal asks for.
A worked example
Suppose your title is Text-message reminders and medication adherence in adults with type 2 diabetes: a randomized controlled trial. Assume, for this example, that the trial ran in primary care clinics and the primary outcome was glycated hemoglobin (HbA1c). Here is how the candidates sort out.
| Candidate | Decision | Reason |
|---|---|---|
| text message | Drop | Already in the title |
| medication adherence | Drop | Already in the title |
| diabetes | Drop | Too broad, and the title is more specific |
| SMS | Spell out as "short message service" | Abbreviation; Short Message Service is an entry term under the MeSH heading Text Messaging |
| mobile health | Keep | Common search term missing from the title; an entry term under the MeSH heading Telemedicine |
| glycated hemoglobin | Keep | Primary outcome, missing from the title; also the MeSH heading |
| primary care | Keep | Setting, missing from the title |
| Telemedicine | Keep only if the journal asks for MeSH headings | Broader heading whose entry terms include mHealth and telehealth |
The resulting list is short: mobile health; short message service; glycated hemoglobin; primary care. Every term adds a route the title does not. Step 6 then sends you back to the abstract to check that it names the primary care setting and spells out glycated hemoglobin before using HbA1c, so the two core concepts are in the text that indexing reads.
Abbreviations, spelling and other edge cases
Established abbreviations. Some abbreviations are how a field actually searches, and MeSH lists many of them as entry terms: "Hb A1c" is one of the entry terms under Glycated Hemoglobin. If the journal allows abbreviations, a sound pattern is to use the spelled-out term as the keyword and define the abbreviation in the abstract.
British and American spelling. Follow the journal's house style. PubMed's search may add the other spelling for you, so in PubMed a spelling variant is a weaker use of a keyword than a genuine synonym. Other databases may handle spelling differently.
Terms newer than the vocabulary. Because MeSH is revised annually, a new technique or condition may not have a heading yet. Use the term your field uses, and add the closest broader heading if the journal wants MeSH.
Classification lists in the submission system. Some journals ask you to pick subject categories from a fixed list as well as typing keywords, and Editorial Manager's documentation describes the same classification list being used both to categorize manuscripts and to identify people's areas of expertise. Choose the most specific categories that genuinely apply. If the journal also invites reviewer suggestions, keywords do not replace a careful list; our guide to suggesting reviewers ethically covers what is acceptable.
Borrowed popular terms. A fashionable term your paper does not address misleads both readers and the systems that match your manuscript to reviewers. Leave it out.
Before you enter your keywords
- Number and format match the journal's instructions
- No keyword repeats a word already in the title, unless the journal asks for it
- Each MeSH heading confirmed in the MeSH Browser, not copied straight from MeSH on Demand
- No single broad words and no ambiguous abbreviations
- Each term searched in PubMed and Google Scholar, with first results that look like your field
- The most important terms also appear in the title or abstract
- Every keyword describes something the paper actually reports
Where discoverability is really decided
The words that decide whether your paper is found are mostly in your title and abstract. The keyword list fills the gaps they leave, so if you only have time for one step, make it step 6.
Whichever journal you choose, read its instructions for authors before you settle the list. Directive Publications keeps its guidance for authors on the for authors page, and you can submit your manuscript there when it is ready.
Frequently asked questions
Should keywords repeat words already in the title?
Usually not. In PubMed, author keywords are searched under the same Title/Abstract tag as your title, so a keyword that repeats a title word adds no new way to find the paper there. Use the space for synonyms, outcomes, settings and methods the title leaves out, unless your journal's instructions ask for something different.
What are MeSH terms and should I use them as keywords?
MeSH (Medical Subject Headings) is the controlled vocabulary the US National Library of Medicine uses to index MEDLINE citations in PubMed. For journals indexed for MEDLINE, NLM assigns MeSH terms itself through automated indexing based on the title and abstract, so your keyword list does not set them. A MeSH heading is still a good keyword when the journal asks for one, or when its wording differs from the words in your title and abstract.
How does Google Scholar decide which papers to show first?
Google Scholar says it aims to rank documents the way researchers do, weighing the full text, where the paper was published, who wrote it, and how often and how recently it has been cited. It does not single out the keyword field. The part you control at submission is whether the terms people search for appear naturally in your title, abstract and text.
Can I use abbreviations as keywords?
Only when the abbreviation is firmly established and unambiguous in your field, and the journal allows it. Many abbreviations have several meanings across medicine and the laboratory sciences, so the spelled-out term is usually the safer keyword. Define the abbreviation in the abstract too, because the abstract is what indexers and search engines read.
Do journals use keywords to select peer reviewers?
Often, as one input. Editorial Manager's help documentation, for example, says classifications and keywords enable the system to help match editors and reviewers to new submissions. The system suggests matches and the editor decides whom to invite, so precise keywords about your method and subfield make a good match more likely.