How to Write a Literature Review

A literature review is one of the most commonly assigned — and most commonly misjudged — sections of a research paper or thesis. Done well, it does real intellectual work: it shows a reader where the study sits within existing knowledge, what question remains genuinely open, and why the present work is justified. Done poorly, it reads as a list of source summaries with no argument connecting them, which is the single most common weakness reviewers and thesis supervisors flag in early drafts.

This guide covers the standard narrative literature review written as a section of a research paper, thesis, or dissertation — its purpose, how to structure it, and the craft skill (synthesis, not summary) that separates a strong review from a weak one. It is not about the standalone systematic review as its own publication type, with a registered protocol and a PRISMA-mandated search strategy; for that distinction, and for when each review type is the right (and correctly labeled) choice, see CASRAI’s comparison of literature review vs. systematic review vs. scoping review. It also assumes you already know where a literature review sits inside a manuscript’s overall structure and how to draft prose generally — for that, see CASRAI’s guides to how to outline a research paper and how to write a research paper. This page picks up where those leave off: how to actually write the literature review section itself.

What a Literature Review Is For

A literature review in a research paper or thesis does three things at once, and losing sight of any one of them is a common source of a weak draft:

It situates the study within existing knowledge. A reader — a reviewer, an editor, a thesis committee member — needs to understand what is already established in the field before they can judge what a new study adds. The review is where that groundwork gets laid.
It identifies a genuine gap or open question. The review has to build, cumulatively, toward the specific problem the rest of the paper addresses — not just describe the field in general, but narrow toward exactly what isn’t yet known, resolved, or agreed on.
It establishes the author’s command of the field. Particularly in a thesis or dissertation, the literature review chapter is also where a committee assesses whether the author actually understands the scholarly conversation they’re joining — not just whether they can list relevant sources.

All three purposes point toward the same conclusion: a literature review that only describes what other people have written, without building toward the specific gap or question the current work addresses, has not done its job — regardless of how many sources it cites or how accurately it describes each one.

Where the Literature Review Fits

How much space a literature review gets, and where it sits, varies by manuscript type:

In a standard IMRaD journal article, the literature review is usually folded into the back half of the Introduction — a few paragraphs that move from broad context to the specific gap the study fills, rather than a separately labeled section. See CASRAI’s guide to the IMRaD structure for how the Introduction is typically organized around this narrowing movement.
In some fields — computer science and engineering particularly — a dedicated, explicitly labeled “Related Work” or “Background” section is conventional, positioned either right after the Introduction or just before the Methods, and organized to position the current work against specific prior approaches.
In a thesis or dissertation, the literature review is typically its own full chapter — substantially longer and more exhaustive than a journal article’s, because it is doing double duty as both scholarly grounding and a demonstration of the author’s command of the field to a committee. See CASRAI’s guide to what a dissertation is for how the literature review chapter fits into a thesis’s overall structure.

Regardless of length or placement, the underlying craft — organizing sources into an argument rather than a list, and synthesizing rather than summarizing — is the same. The rest of this guide covers that craft directly.

Structuring a Literature Review: Thematic, Chronological, and Methodological

The most common structural mistake in a literature review draft is organizing it source-by-source — one paragraph per study, in whatever order they were found — instead of around an actual organizing principle. University writing centers and library research guides consistently teach three main organizing patterns, and a strong review usually foregrounds one while drawing on the others where useful:

Thematic. Sources are grouped around central themes, concepts, or points of debate rather than by publication date. This is generally the strongest default for a complex topic with multiple dimensions, because it lets the review move fluidly between older and newer sources within each theme, according to the point being made, rather than being locked into chronological order.
Chronological. Sources are arranged to trace how thinking, methods, or findings on a topic have developed over time. This organization is the right choice specifically when the point of the review is to show an evolution — a field’s methods getting more rigorous, a debate shifting, an early consensus being overturned — not simply as a default way to sequence sources.
Methodological. Sources are grouped by research approach — qualitative versus quantitative, experimental versus observational, the specific instruments or models used — rather than by what they found. This fits when methodology itself is a driver of divergent findings in the field, and the review needs to explain why studies disagree by pointing to how they were conducted.

These patterns aren’t mutually exclusive — a primarily thematic review will often use a brief chronological note within a theme to show how that specific idea developed. The failure mode to avoid is defaulting to a fourth, unnamed pattern: a source-by-source list with no organizing principle at all, which reads as an annotated bibliography rather than a literature review (see the distinction below).

Synthesis vs. Summary: The Difference That Actually Matters

This is the single most consequential craft distinction in literature review writing, and the one most often missing from early drafts — including in otherwise well-researched student and early-career manuscripts.

Summary describes what an individual source says, one source at a time: “Smith (2019) found X. Jones (2021) found Y. Lee (2022) found Z.” Each sentence or paragraph is anchored to a single study, and the connections between studies — if any — are left for the reader to infer. This is exactly the structure of an annotated bibliography, and it is a legitimate genre in its own right, but it is not a literature review.

Synthesis reorganizes that same information around the author’s own argument or theme, rather than around each individual source. A synthesized passage groups Smith, Jones, and Lee together because they bear on the same question, and states explicitly how they relate to one another: where they agree, where they diverge, whether a later study built on, contradicted, or narrowed an earlier one, and what that pattern implies about what is and isn’t yet known. The citations are still there — synthesis does not mean fewer sources or less rigor — but the sentence-level organizing unit shifts from “this source says” to “this idea, and here is who supports, complicates, or contests it.”

A useful self-check while drafting: read a paragraph back and ask whether it could be reordered into a bulleted, one-line-per-source list without losing anything. If the answer is yes, it’s summary, not synthesis — the paragraph isn’t doing anything a list couldn’t do. A synthesized paragraph resists that reordering, because its structure is built around the relationship between sources, not around the sources themselves.

A Practical Technique: The Synthesis Matrix

Moving from a pile of read sources to a synthesized draft is a genuine drafting bottleneck, and a technique widely taught in academic library and writing-center guidance is worth using deliberately rather than skipping straight from reading to drafting: the synthesis matrix. The mechanics are simple — build a table (a spreadsheet works fine) with one row per theme or concept the review will address, and one column per source. As each source is read (or reread), its relevant notes, findings, or quotations go into the cell where that source’s row and column intersect.

The payoff comes once the matrix is filled in: reading across a single row shows, at a glance, everything every source says about one theme — which is precisely the organizing unit a synthesized paragraph needs, and precisely the view a source-by-source reading of notes does not give. Gaps become visible the same way: a row with only one or two filled cells is either a genuinely thin area of the literature (a candidate for the gap statement the review is building toward) or a sign that more sources need to be found for that theme before drafting that section.

Citation Management for a Literature Review

A literature review typically draws on more sources, tracked over a longer working period, than any other section of a manuscript — which makes citation-management discipline more consequential here than elsewhere in the paper. A few practicalities worth setting up before drafting, not after:

Use a reference manager from the start. Zotero, EndNote, and Mendeley are the three most widely used reference managers in academic research; all three let a source be tagged, grouped into collections or folders, and annotated, and all three generate a correctly formatted in-text citation and reference list in whatever style a target journal or institution requires. Tag or file sources by theme as they’re added — this does double duty as a lightweight synthesis matrix and keeps the eventual reference list accurate without a separate manual pass. See CASRAI’s guide to citation styles and referencing formats for how these tools apply Citation Style Language (CSL) definitions to structured metadata to produce a specific style’s formatting automatically.
Attach the citation the moment a claim is drafted, not as a separate formatting pass at the end — the same discipline CASRAI’s guide to writing a research paper recommends for the manuscript generally applies with extra force to a literature review, where the density of cited claims per paragraph is highest and a missed or misattributed citation is easiest to lose track of.
Use a citation-mapping tool for discovery and gap-checking, not just formatting. Tools like Research Rabbit and Litmaps build a visual citation graph from a small set of seed papers, surfacing related, foundational, and citing work that a keyword search alone can miss — useful specifically for confirming a literature review’s search hasn’t missed an obvious cluster of related work, even though a narrative review, unlike a systematic review, has no formal exhaustive-search requirement.
Keep the final source count proportionate. There is no fixed rule for how many sources a literature review should cite — it depends on the depth of the field and whether it’s a journal article’s background section or a full thesis chapter — but a review with very few sources for a well-established topic, or an unreviewed flood of sources with no filtering, are both signs the selection process itself needs a second pass before the writing does.

Common Pitfalls

Writing an annotated bibliography and calling it a literature review. A source-by-source list of what each study found, with no explicit connective argument between them, is the single most common structural weakness — see the synthesis-vs-summary section above.
No organizing principle at all. Sources presented in whatever order they were found, rather than deliberately arranged thematically, chronologically, or methodologically.
A review that never narrows to a gap. Describing the field broadly and comprehensively, but never arriving at the specific open question the rest of the paper addresses, leaves the reader to guess why the review was necessary.
Treating citation as a final formatting pass. Planning to “add citations later” increases the risk of losing track of which claim came from which source — a particular risk in a section with this many cited claims per paragraph.
Confusing “literature review” with “systematic review.” Labeling a narrative, author-selected review as “systematic” without doing the protocol-driven, exhaustive-search work that label requires is a common and avoidable cause of reviewer pushback — see CASRAI’s literature review vs. systematic review vs. scoping review comparison for the full distinction.

Frequently Asked Questions

What’s the difference between a literature review and a systematic review?

A literature (narrative) review has no fixed protocol — the author selects sources based on their own judgment, and the search doesn’t need to be exhaustive or reproducible. A systematic review answers a narrow, pre-defined question via a registered protocol, an exhaustive and reproducible search, and a formal risk-of-bias appraisal, reported against the PRISMA 2020 checklist. See CASRAI’s full comparison of literature review, systematic review, and scoping review.

How long should a literature review section be?

There’s no fixed length — it depends heavily on genre. A journal article’s literature review is often just a few paragraphs folded into the Introduction; a thesis or dissertation literature review is typically a full chapter, often the longest single chapter in the document, because it also has to demonstrate the author’s command of the field to a committee. Check the target journal’s word limits or the institution’s thesis guidelines rather than assuming a fixed page count.

Should a literature review be organized by source or by theme?

By theme (or chronologically, or methodologically) — not source-by-source. Organizing paragraph-by-paragraph around individual sources produces something closer to an annotated bibliography than a literature review. See the structuring and synthesis-vs-summary sections above.

How many sources should a literature review cite?

There’s no universal number — it depends on how established the field is and whether the review is a journal article’s background section or a full thesis chapter. What matters more than a target count is whether the sources actually cover the field’s major themes and debates; a synthesis matrix (see above) is a practical way to check for thin coverage before drafting.

Can I use a reference manager like Zotero or EndNote to help write a literature review?

Yes — tagging and grouping sources by theme in a reference manager as they’re collected is a practical way to keep a literature review’s source base organized from the start, and all major reference managers generate a correctly formatted reference list in whatever citation style is required. See CASRAI’s guide to citation styles and referencing formats for how that formatting actually works.

Can I use AI tools to help write a literature review?

Policies vary by publisher and institution, and are changing quickly — most major publishers permit AI-assisted drafting or editing if disclosed, but none permit listing an AI tool as an author, and using an AI tool to fabricate or misrepresent what a source actually says is a distinct integrity problem regardless of disclosure. See CASRAI’s guide to generative AI policies across publishers before assuming a specific use is permitted undisclosed.

Related CASRAI Resources

Scholarly Writing pillar — the full manuscript-preparation and publishing-craft cluster this guide belongs to
Literature Review vs. Systematic Review vs. Scoping Review — the full distinction between review types
How to Write a Research Paper — the drafting, style, and revision process this guide’s literature review section fits inside
IMRaD Structure / How to Outline a Research Paper
What Is a Dissertation? — where a full literature review chapter fits in a thesis
Citation Styles and Referencing Formats
Research Rabbit: Citation-Mapping Tool Guide
Litmaps: Citation Mapping and Literature Monitoring Tool
PRISMA and Systematic Review Methodology