Quick answer
A literature review (narrative review) is an interpretive synthesis — the author selects and argues from prior work to motivate a question or position. A systematic review is a research method: a pre-registered protocol, exhaustive reproducible search, dual screening, and formal risk-of-bias assessment, reported under PRISMA 2020. Use a narrative review for broad or theoretical questions and thesis chapters; use a systematic review when a focused clinical or policy question demands auditable completeness.
The two genres answer different questions
Both review types synthesize existing research, which is where the confusion starts. The deeper difference is epistemic. A narrative literature review derives its authority from the author's expertise: you are trusted to know which papers matter, and your selection bias is an accepted feature, disclosed through voice and framing. A systematic review derives its authority from its method: the protocol, the search strings, and the screening log are designed so that the author's judgment touches the study set as little as possible. One is an argument; the other is an audit.
That difference cascades into everything else — how you search, how long it takes, how many people you need, and which one your examiner or target journal actually expects. The table below names the nine dimensions where the genres genuinely diverge.
Literature review vs systematic review: side by side
| Dimension | Literature review (narrative) | Systematic review |
|---|---|---|
| Purpose | Build an argument: synthesize and interpret prior work to motivate a research question, frame a debate, or take a position. | Answer one focused, pre-specified question (often PICO-framed) by finding and appraising every study that meets explicit criteria. |
| Protocol | None required. The author decides scope, sources, and organization as the review develops. | Written and pre-registered before searching (PROSPERO for health topics, OSF elsewhere). Deviations must be declared and justified. |
| Reproducibility | Not reproducible by design — two experts reviewing the same topic will legitimately select and weight different papers. | A core requirement. Another team following your documented search strings, databases, and criteria should retrieve the same study set. |
| Search strategy | Purposive: seed papers, forward and backward citation chasing, expert knowledge of the field. No completeness claim. | Exhaustive: named databases (e.g. MEDLINE, Embase, Scopus, Web of Science), full Boolean strings reported verbatim, plus grey literature and citation searching. |
| Screening | Single author selects what is relevant; no formal record of what was excluded or why. | Dual independent screening of titles/abstracts then full texts, with conflicts resolved by a third reviewer; every exclusion at full-text stage gets a recorded reason. |
| Bias handling | Author bias is accepted and disclosed through voice and framing — the reader knows it's one expert's reading. | Formal risk-of-bias assessment per included study (RoB 2, ROBINS-I, Newcastle–Ottawa), plus publication-bias checks (funnel plots, Egger's test) where pooling occurs. |
| Typical length | Flexible: 1,500-word journal mini-review to a 10,000–15,000-word dissertation chapter. | 5,000–12,000 words of text plus a PRISMA flow diagram, study-characteristics tables, and often lengthy supplementary appendices. |
| Time required | Weeks to ~3 months for one author working part-time on it. | 6–18 months with a team of two or more; Cochrane estimates commonly land near 12 months from protocol to submission. |
| When it's expected | Thesis/dissertation chapter 2, journal-article introductions, grant background sections, theory and position papers. | Clinical and policy questions, evidence for practice guidelines, health-technology assessments, and journals (BMJ, Cochrane, JBI) that mandate PRISMA reporting. |
These are reporting-guideline-backed conventions, not preferences: the systematic-review column tracks PRISMA 2020 and the Cochrane Handbook; the narrative column tracks what examiners and journal editors accept in practice. For worked narrative-review structures by field, see our literature review examples.
Which one should you write? A four-question flowchart
Work through these in order. Each question is a real gate — the most expensive mistake is answering "yes" aspirationally at question 3 and discovering eight months in that you never had the team or the time.
- 1
Is your question focused enough to state in PICO form (population, intervention, comparator, outcome)?
No → narrative literature review. Broad, exploratory, or theoretical questions can't be answered by exhaustive search; they need interpretive synthesis.
Yes → continue.
- 2
Do you need the result to inform clinical practice, policy, or a guideline — where missing studies would change the answer?
No → a narrative review (or a scoping review, if you want breadth with some rigor) is usually sufficient and 10× faster.
Yes → continue.
- 3
Do you have 6–18 months, at least two reviewers, and access to the major databases?
No → consider a rapid review (a streamlined systematic method with declared shortcuts) or scale back to a scoping review. Don't run a one-person 'systematic' review — examiners and editors will spot it.
Yes → systematic review. Register the protocol before you run a single search.
- 4
Are the included studies similar enough in design and outcome measures to pool statistically?
No → systematic review with narrative synthesis (SWiM guideline applies).
Yes → systematic review with meta-analysis.
If the flowchart lands you on the narrative side, your next stops are the how to write a literature review guide and a working literature review outline — structure decisions made early save weeks of restructuring later.
PRISMA 2020 essentials
PRISMA 2020 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) is the reporting standard nearly every health-science journal — and increasingly education, software engineering, and management journals — requires for anything labeled systematic. The non-negotiables:
- Protocol and registration. Write the protocol before searching; register it on PROSPERO (health) or OSF (everything else). The registration timestamp is your proof that eligibility criteria weren't bent to fit the results.
- The 27-item checklist. Submitted alongside the manuscript; editors check it line by line. Items cover eligibility criteria, information sources, full search strings, the selection and data-collection process, risk-of-bias methods, and synthesis methods.
- The flow diagram. The four-stage funnel — records identified → screened → assessed for eligibility → included — with counts and exclusion reasons at every stage. It is the single most-inspected figure in the genre.
- Reproducible search reporting. Every database named, every Boolean string reproduced verbatim (usually in an appendix), search dates stated, and supplementary methods — including forward citation searching of included studies — documented.
- Risk-of-bias and certainty assessment. A named instrument per study design (RoB 2 for randomized trials, ROBINS-I for non-randomized studies) and, increasingly, GRADE ratings for the certainty of each outcome-level conclusion.
The middle ground: scoping reviews and meta-analyses
The narrative/systematic binary hides a useful spectrum. Two in-between forms matter most:
Scoping reviews borrow the systematic machinery — protocol, structured multi-database search, dual screening — but ask a mapping question ("what evidence exists, of what types, with what gaps?") instead of an effectiveness question, and they normally skip risk-of-bias assessment. They are reported under PRISMA-ScR. A scoping review is the right call when the field is broad or messily defined, or when you need to demonstrate that a full systematic review is (or isn't) feasible before committing a team to one. It is also the honest upgrade path for a solo researcher who wants more rigor than a narrative review but can't staff dual extraction and bias assessment.
Meta-analyses sit on the far end: the statistical pooling of quantitative results across the studies a systematic review identified. A meta-analysis is not a separate review type — it is an optional final step, valid only when the included studies are similar enough in design, population, and outcome measures. Heterogeneous evidence gets narrative synthesis instead (reported under the SWiM guideline). Remember the dependency direction: a meta-analysis inherits every weakness of the search beneath it. Pooling 12 studies impressively does not fix having missed 8.
Common mistakes (and how examiners catch them)
Calling a narrative review "systematic"
The single most common error examiners and editors flag. If you didn't pre-register a protocol, search multiple named databases with reported strings, dual-screen, and assess risk of bias, the word "systematic" doesn't apply — say "narrative review" or "literature review" instead. Mislabeling invites desk rejection and, in a viva, a very uncomfortable line of questioning.
Skipping pre-registration
Registering after the searches are done (or not at all) defeats the purpose — the protocol exists to stop you from quietly changing eligibility criteria once you've seen the results. PROSPERO registration is free and most health-science journals now check the registration date against your search dates.
One person doing the screening
Single-reviewer screening misses roughly 8–13% of eligible studies in methodological studies of the process. If you genuinely can't recruit a second reviewer, declare single-screening as a limitation explicitly — don't present it as dual screening.
Searching only one database (or only Google Scholar)
Google Scholar is excellent for citation chasing but irreproducible as a primary systematic search — results vary by user and date, and it can't export full result sets. Use it as a supplementary source on top of at least two structured databases.
Confusing comprehensiveness with synthesis
A systematic review that ends in a table of 47 studies with no analytic comparison is as weak as a narrative review with no argument. Both genres are judged on synthesis — the systematic one just has to document how the evidence was assembled first.
How citation mapping supports both review types
Citation mapping earns its place at different stages of each genre.
For narrative reviews: seed-paper expansion. The fastest way to scope a field is to map the citation footprints of 3–5 seed authors whose names recur in everything you've read. A geographic citation map shows you, in seconds, which countries and institutions are actually carrying the conversation — including the research traditions (East Asian, Latin American, continental European) that English-language keyword search systematically under-surfaces. Those clusters become the named themes your review is organized around, and the empty regions become candidate gap statements. The full workflow is in the citation map for literature review guide.
For systematic reviews: documenting forward-citation coverage. PRISMA 2020 expects supplementary search methods alongside the database queries, and forward citation searching of included studies is the standard one — for each included study, you walk every paper that cites it, because new studies cite their predecessors even when keyword indexing misses them. A citation map across the included set does two jobs here: it operationalizes the chase (every citing paper, grouped by institution and country, rather than an unpaginated Google Scholar list), and it documents the geographic and language reach of that supplementary search — evidence that belongs verbatim in your limitations section when, say, Chinese- or Spanish-language clinical literature is visibly thin in your citing-paper map.
Either way, the entry point is the same: paste a Google Scholar profile into the search page and the geographic map of citing institutions renders in seconds.
Frequently asked questions
What is the main difference between a literature review and a systematic review?+
A literature review (narrative review) is an interpretive synthesis: the author selects, organizes, and argues from prior work, and the selection is openly judgment-based. A systematic review is a research method in its own right: a pre-registered protocol, exhaustive reproducible search, dual screening, and formal risk-of-bias assessment, all reported under PRISMA 2020. The narrative review answers 'what should we make of this field?'; the systematic review answers one focused question with auditable completeness.
Is a systematic review better than a literature review?+
No — they do different jobs. A systematic review is better when missing studies would change the answer (clinical effectiveness, policy decisions, guideline development). A narrative review is better for broad, theoretical, or fast-moving questions, for motivating a thesis or grant, and whenever you need expert interpretation rather than exhaustive enumeration. Most PhD dissertations correctly use a narrative review for chapter 2; forcing a systematic protocol onto an exploratory question wastes 6–12 months.
How long does a systematic review take compared to a literature review?+
Plan 6–18 months for a systematic review with a team of at least two reviewers — protocol writing, searching, dual screening of hundreds to thousands of records, data extraction, risk-of-bias assessment, and synthesis each take weeks. A narrative literature review typically takes a few weeks to about three months for a single author. Rapid reviews split the difference (2–6 months) by declaring methodological shortcuts up front.
What is a narrative review vs a systematic review?+
'Narrative review' is the precise name for the traditional literature review: a synthesis whose source selection is purposive and whose authority comes from the author's expertise. A systematic review replaces that judgment-based selection with a pre-specified, reproducible pipeline. The terms 'literature review' and 'narrative review' are near-synonyms; 'systematic review' is not interchangeable with either, and using it loosely is a common cause of desk rejection.
Can one person do a systematic review?+
Technically yes, but it falls short of methodological expectations: PRISMA 2020 and Cochrane both assume at least two independent screeners, because single-reviewer screening demonstrably misses eligible studies. A solo researcher (e.g. a master's student) should either recruit a second screener for the screening stage only, run a scoping review instead, or explicitly declare single-reviewer screening as a limitation. Examiners accept an honest limitation far more readily than an inflated label.
Does a systematic review have to include a meta-analysis?+
No. A meta-analysis is the optional statistical step of pooling quantitative results across studies, and it is only valid when studies are similar enough in design, population, and outcome measures. Many systematic reviews end in narrative synthesis (reported under the SWiM guideline) because the evidence is too heterogeneous to pool. The reverse also holds: a meta-analysis is only as good as the systematic search underneath it.
What is a scoping review, and when should I choose it instead?+
A scoping review uses systematic-style searching and screening but asks a mapping question — 'what evidence exists and what are its characteristics?' — rather than an effectiveness question, and it usually skips risk-of-bias assessment. It is reported under PRISMA-ScR. Choose it when the field is broad or poorly defined, when you want to catalogue study types and gaps before committing to a full systematic review, or when you have systematic-review-level rigor ambitions but a narrative-review-sized question.
How does citation mapping help with either review type?+
For a narrative review, citation maps accelerate seed-paper expansion: map the citation footprint of 3–5 key authors and you surface the clusters, labs, and regions you should be reading. For a systematic review, forward-citation searching of included studies is a PRISMA-recommended supplementary search — and a geographic citation map documents the coverage (and the language/region blind spots) of that step, which belongs in your limitations section.
Keep going
- Literature review examples — five worked cases by field plus seven format samples, including a PRISMA systematic review sample.
- How to write a literature review — the step-by-step narrative-review process.
- Literature review outline — ready-to-adapt structures by format.
- Citation map for literature review — the geographic-mapping workflow in depth.