A review article is a standalone, peer-reviewed publication whose entire content is a synthesis of existing literature on a defined question or field — not a section embedded in an original-research paper, and, in most cases, not a systematic review with a registered protocol. Journals across nearly every discipline publish a dedicated review-article category (Annual Reviews’ entire portfolio is built on the format; most clinical, life-science, and social-science journals carry a regular review section), and many actively solicit them because a good review is heavily cited and widely read. This guide covers how to plan, structure, and write a narrative review for submission as its own manuscript.
This is a different task from writing the literature-review section of an original-research paper or thesis — for that, see CASRAI’s guide to how to write a literature review for a research paper. It is also a different task, with different rules, from a systematic review. Where those two overlap and diverge is covered in full in CASRAI’s comparison of literature review vs. systematic review vs. scoping review — read that first if you are not yet sure which type your project actually is, since submitting a narrative review as if it were systematic (or vice versa) is a common and avoidable cause of desk rejection.
Narrative Review vs. Systematic Review: Why the Distinction Matters Here
A narrative (traditional) review article does not require a registered protocol, a PROSPERO registration, or adherence to PRISMA 2020 reporting items — those requirements, and the formal search-and-screening methodology behind them, belong to systematic reviews and their close relatives (scoping reviews, meta-analyses). A narrative review instead relies on the author’s expert judgment to select, weigh, and synthesize the literature, and its organizing structure is typically conceptual or thematic rather than a documented, reproducible search protocol. CASRAI covers that fully rigorous, protocol-driven methodology separately in PRISMA and systematic review methodology — if your research question calls for an exhaustive, bias-minimized, reproducible synthesis (for example, to inform a clinical guideline or a policy decision), a systematic review is very likely the more defensible choice, and PROSPERO explicitly does not accept narrative or scoping reviews for registration, only systematic reviews and their variants.
That does not make a narrative review a lesser or less rigorous format — it makes it a different one, suited to different questions. A narrative review is usually the right choice when the goal is to survey a broad or fast-moving field, trace how thinking on a topic has developed, synthesize across disciplines that a systematic protocol couldn’t cleanly span, or offer expert interpretation and a forward-looking argument rather than an exhaustive accounting of every eligible study. What a narrative review still owes the reader, however, is a genuine synthesis — connecting, comparing, and critically evaluating sources toward an argument — not a list of source summaries strung together with no throughline. Reviewers routinely flag exactly that weakness, and it is the most common reason a submitted narrative review reads as a literature summary rather than a scholarly contribution.
Before You Start Writing
Define a real question, not just a topic
“Recent advances in X” is a topic, not a question. A publishable review article argues something: that the field’s dominant framework has a specific weakness, that findings across two subfields conflict in an identifiable way, that a consensus has emerged (or hasn’t) on a defined issue. Framing a specific question or thesis before drafting is what turns a review from a summary into a contribution — and it is the single factor editors and reviewers cite most often when distinguishing a review they’ll accept from one they’ll reject as “just a summary.”
Check whether the journal accepts unsolicited reviews
Many journals that carry a review section only publish invited reviews, commissioned by an editor from a recognized expert. Others accept unsolicited proposals or full manuscripts. Before drafting, check the specific journal’s author guidelines for its review-article policy, word-count ceiling (commonly 3,000–6,000 words, though this varies widely by discipline and journal), and whether a pre-submission inquiry or abstract is expected. This is a submission-mechanics question distinct from the writing task itself; see CASRAI’s cover letter for journal submission entry for what a submission letter typically needs to state.
Scope the literature realistically
A narrative review’s lack of a mandated search protocol is a source of flexibility, not an excuse to skip a systematic search entirely. Reviewers and editors expect you to describe, at least briefly, how you identified the literature you’re drawing on — which databases, what date range, what you deliberately excluded and why. The SANRA scale (Scale for the Assessment of Narrative Review Articles, Baethge, Goldbeck-Wood & Mertens, 2019, published in Research Integrity and Peer Review) is the most widely used quality framework specifically built for this format, and its six items are a practical checklist while you plan: justification of the review’s importance, a clearly stated aim or question, an adequately described literature-search approach, appropriate referencing, sound scientific reasoning, and appropriate presentation of data and information. Several journals now ask authors to self-assess against SANRA before submission, or ask reviewers to score against it.
Structuring the Manuscript
Unlike a systematic review, a narrative review has no single mandated section structure (there is no narrative-review equivalent of the PRISMA checklist or flow diagram). That flexibility is real, but most published narrative reviews converge on a recognizable shape:
- Title and abstract. The title should signal both the topic and the review’s angle or argument, not just the subject area. Most journals expect a structured or semi-structured abstract summarizing the review’s scope, its main synthesized conclusions, and its significance — check the target journal’s abstract requirements specifically, since these vary more for reviews than for original research.
- Introduction. Typically brief (a few paragraphs): establishes why the topic matters now, states the review’s specific aim or question, and previews how the review is organized. This is where the review earns its right to exist as a contribution rather than a restatement of settled knowledge.
- Body, organized by theme, chronology, or methodology — not by source. This is the section most narrative reviews get wrong in early drafts. Organizing by individual study (“Smith et al. found X; Jones et al. found Y”) produces a summary, not a synthesis. Organizing by theme, subtopic, mechanism, or debate — and using individual studies as supporting evidence within that structure — produces the argument-driven synthesis reviewers expect. For a broad topic, subdividing the body into clearly labeled subsections (by mechanism, population, time period, or methodological approach, depending on the field) both aids readability and forces a more disciplined organizing logic than a single continuous narrative.
- Critical evaluation, not just description. A narrative review is expected to weigh evidence, not just report it — noting where studies conflict, where evidence is strong versus preliminary, and where methodological limitations in the underlying literature limit what can be concluded. This is the “sound scientific reasoning” SANRA item, and it is what separates a review with real editorial value from an annotated bibliography.
- Conclusion and future directions. Should return explicitly to the question posed in the introduction, state what the synthesis actually resolves or clarifies, and typically identify open questions or needed future work — a common and expected close for a review article, since part of its value to the field is pointing toward what isn’t yet known.
- References. Review articles are often reference-heavy (commonly 50–150+ citations depending on scope and discipline) precisely because comprehensiveness of coverage, even without a formal search protocol, is still part of what makes a narrative review credible. Reference-management practicalities are the same as for any manuscript — see CASRAI’s guide to writing a literature review section for citation-management conventions that apply equally here.
Common Weaknesses Editors and Reviewers Flag
- Summary without synthesis. The most frequently cited weakness: a review that restates what each source found, in source order, without building an argument that connects them.
- No stated methodology at all. Even without a PRISMA protocol, a narrative review needs some transparency about scope and search approach; omitting this entirely undermines credibility and is explicitly penalized under SANRA’s literature-search item.
- Undisclosed selectivity or one-sidedness. Because a narrative review’s inclusion criteria are the author’s judgment rather than a documented protocol, reviewers watch closely for whether contrary evidence and competing interpretations are represented fairly, not just evidence supporting the author’s own prior position.
- Scope creep. A review that tries to cover too broad a field superficially, rather than a narrower question in real depth, tends to satisfy neither breadth nor depth well. Narrowing the stated question before drafting is the most effective fix.
- Mislabeling. Calling a narrative review a “systematic review” (or the reverse) in the title or abstract, without meeting that format’s actual methodological requirements, is flagged by editors and, increasingly, by editorial software that checks reporting-guideline compliance against the stated review type. Use the terminology that matches what you actually did — see the literature review vs. systematic review vs. scoping review comparison for how the major review types are distinguished and labeled correctly.
Related CASRAI Resources
CASRAI’s narrative review and systematic review dictionary entries give the operational definitions behind the distinctions on this page. For the fully protocol-driven alternative, see PRISMA and systematic review methodology. For quantitative pooling of results across studies (a different task from either kind of narrative synthesis), see the systematic review vs. meta-analysis comparison and the meta-analysis dictionary entry. For the literature-review section embedded within an original-research manuscript rather than a standalone review article, see how to write a literature review for a research paper.
Frequently Asked Questions
Does a narrative review article need ethics approval or a registered protocol?
No. Ethics approval applies to primary research involving human subjects, animal subjects, or identifiable data, not to a synthesis of already-published literature. Protocol registration (PROSPERO) is specific to systematic reviews and their variants; PROSPERO does not accept narrative review registrations.
How long should a review article be?
There is no universal figure — it depends entirely on the target journal and field, commonly in the range of 3,000 to 6,000 words for a focused review, though comprehensive reviews in fields like Annual Reviews titles run considerably longer. Always check the specific journal’s word-count policy for its review-article category before drafting.
Can a narrative review be a solo-authored piece, or does it need to be a team effort?
Either is common. Narrative reviews are frequently solo-authored, especially when commissioned from a single recognized expert, but multi-author reviews are equally normal, particularly for broad or interdisciplinary topics where no one author reasonably covers the full scope.
Is a narrative review considered original research for CV or promotion purposes?
Policies vary by institution and field. A review article is a distinct, real, peer-reviewed publication type and is typically listed as such, but whether it counts toward criteria specifically requiring “original research” output depends on the evaluating body’s own policy — check with your institution or funder rather than assuming either way.
What’s the difference between a narrative review and a scoping review?
A scoping review sits methodologically between a narrative and a systematic review: like a systematic review, it uses a documented, reproducible search-and-screening protocol (often following PRISMA-ScR reporting), but like a narrative review, its purpose is typically to map the breadth of a field or identify gaps rather than to answer a narrowly defined, quantifiable question. See the scoping review dictionary entry and the three-way comparison linked above for the full distinction.
