Learn

How to write a literature review: a complete guide

What a literature review actually is

A literature review is not a summary of everything that's been written on a topic. This is the most common misunderstanding about the form, and it produces the most common kind of bad literature review: a long parade of paragraph-length summaries of individual sources, one after another, with no argument connecting them.

A literature review is an argument about the state of knowledge in a field. It establishes what is known, what is contested, where the important debates are, what methodological approaches have been used, what gaps exist, and, crucially, why your research question matters in light of all of this. The sources are evidence for that argument, not the subject of it.

The difference sounds subtle but it changes everything about how you write. A summary-based review asks: what does each source say? An argument-based review asks: what do these sources collectively reveal about the field, and what does that mean for my project?

Literature reviews appear in different contexts with slightly different purposes. A standalone literature review is a form of research output in itself, synthesising a field at a particular moment. A dissertation or thesis literature review establishes the intellectual context for your own research question, showing that your question is both important and not already answered. A systematic review is a rigorous method for synthesising evidence on a specific question, often used in healthcare and policy research.

This guide is primarily focused on dissertation and thesis literature reviews, though most of the principles apply broadly.

Before you start searching

The two biggest time-wasters in a literature review are searching without a clear question and searching without an organised system for managing what you find.

Articulate your research question as precisely as possible before you search. A vague question produces a vague search. "What are the effects of X on Y?" is more useful than "X and Y." "How has the understanding of Z changed between 1990 and 2020?" is more useful than "Z." The more specific the question, the more useful the search, and the more manageable the results.

Set up your organisation system before the first result. A folder for PDFs, a system for literature notes, a place to develop synthesis notes, and a way to track what you've searched (which databases, which search terms, which dates). This takes an hour at the start and saves days of confusion later. See the research workflow guide for a detailed approach.

Define the scope. You cannot and should not read everything. Decide in advance: what time period are you covering? What geographic or disciplinary scope? What types of sources are you including (peer-reviewed articles, books, grey literature, policy documents)? Documented scope decisions make the review defensible.

Searching systematically

The key word in "systematic review" is systematic: documented, reproducible, and comprehensive within the defined scope. Even if you're not writing a formal systematic review, bringing some of that rigour to your search will make your literature review stronger.

Use multiple databases. Searching only Google Scholar is not systematic. For most social science and humanities topics, starting with the major databases in your field (JSTOR, Scopus, PsycINFO, ERIC, Web of Science, or discipline-specific equivalents) and combining results is standard practice. For your field's conventions, ask your supervisor or a subject librarian.

Develop a search strategy. Before you search, write down your key terms and their synonyms. If your topic involves "working memory," you might also search for "short-term memory," "cognitive load," and "attention capacity." A documented search strategy means you can repeat the search later and explain it to others.

Use Boolean operators. AND narrows results (memory AND attention returns only results mentioning both). OR broadens them (memory OR recall). NOT excludes (memory NOT computer). Combining these with parentheses gives you precise control: (working memory OR short-term memory) AND (learning OR education) AND (children OR adolescents).

Follow the reference chains. Systematic database searching is one route to sources. Following references in key papers (backwards citation chasing) and finding papers that cite a key paper (forwards citation chasing, using Google Scholar's "Cited by" function) are often more efficient routes to the most important work in an area.

Track everything you search. Which databases, which terms, which date ranges, how many results. This documentation matters if anyone asks how you found your sources, and it prevents you from repeating the same searches three months later.

Reading and taking notes

With a large volume of literature to read, the temptation is to read everything and then synthesise. This is the wrong order. Reading a source without a clear question produces unfocused notes. Reading a source with your research question in mind produces material you can actually use.

Read abstracts first. For each search result, the abstract tells you whether the full text is worth reading. Most results won't be. Be ruthless about inclusion: does this directly address your research question or a key sub-question? If not, it's either a "maybe" for later or excluded.

Take literature notes for every source you include. Title, author, year, what the main argument is, what evidence or methodology is used, what's relevant to your research question, any quotes you might use (with page numbers). Fifteen minutes per source, written immediately after reading. These notes are what your review is built from.

Note the relationship between sources. When you're reading source B and it directly responds to source A, or contradicts it, or builds on it, write that down. These relationships between sources are the substance of a literature review. A review that notes "Smith (2018) argues X; Jones (2020) argues Y" is weaker than one that notes "Jones (2020) directly challenges Smith's (2018) methodology, arguing that..."

Be alert to methodological debates. Many fields have ongoing arguments not just about findings but about how to study a topic. A literature review that ignores these is incomplete. What methods have been used? What are their limitations? Which approaches are currently contested and why?

Organising and synthesising

Once you have notes from enough sources to see patterns, the synthesis work begins. This is the hardest part and the part most students skip.

Organise by theme, not by source. Not "Notes on Smith. Notes on Jones. Notes on Williams." Instead: "What is established about X. What is contested about Y. How Z has been defined and measured. The main methodological approaches."

This reorganisation is the cognitive work of a literature review. You're building the argument from the ground up: identifying what's known, what's debated, what approaches have been tried, where the gaps are. Each theme is a node in the structure of your eventual review.

Write synthesis notes. For each theme you've identified, write a short note: what do the sources collectively say about this? Where do they agree? Where do they diverge? What remains unresolved? These notes are the scaffolding for your review chapters.

Identify the gap. A literature review in a dissertation context is building toward the justification for your own research question. The gap is not a hole where nobody has done any research. It's more often: an unresolved debate where your approach might shed light; a well-studied phenomenon in one context that hasn't been studied in yours; a methodological limitation that your study addresses; a question that the existing literature raises but doesn't answer. Making this gap explicit is what transforms a review from a description of the field into an argument for why your project matters.

Draw a conceptual map. Before you write, try to represent the structure of the field visually: the main debates, who takes which positions, what's settled and what's contested, how different concepts relate. A canvas or whiteboard works well for this. The map doesn't go in your review, but writing it helps you understand the territory before you describe it.

Writing the review

A literature review is an argument written in academic prose. The structure should reflect the argument, not the chronology of your reading or the categories of your database searches.

Start with the argument, not the sources. Your review has a thesis. It might be implicit (this is where the field is and here's what it leaves open) but it should be articulable. Write that thesis in a sentence before you start drafting. Every paragraph should do something to develop or substantiate that thesis.

Use topic sentences. Each paragraph should start with a sentence that states what the paragraph argues, not who you're about to summarise. "The debate about X has centred on two methodological approaches" is a topic sentence. "Smith (2018) argues that..." is the beginning of a summary, not an argument.

Group sources by what they contribute to your argument, not by date or author. "Several studies have found that X (Smith, 2018; Jones, 2020; Williams, 2021)" is more analytical than three separate paragraphs on each study. Group when sources make similar points; distinguish when they diverge.

Write the critical analysis, not just the description. This is the most consistent weakness in student literature reviews. Describing what a source argues is not enough. Evaluating it is: what are the methodological limitations? How does it relate to other work? What does it not account for? Critical evaluation doesn't mean hostile; it means engaged, precise, and analytical.

Integrate sources into your argument. Sources appear in your review in service of a point you're making. The point comes first; the source provides evidence or example. The reverse order (source-led writing, where you say "Smith argues X" and then explain what that means) produces a review that reads as a sequence of source descriptions.

Common mistakes

The annotated bibliography. One paragraph per source, in alphabetical or chronological order, describing each. This is not a literature review. It describes sources but doesn't synthesise them.

The everything-in review. Trying to cover all possibly relevant work produces a review that is both too long and too thin. Depth on the most important material is better than width across everything tangentially related.

Ignoring conflicting evidence. A review that only cites work supporting your eventual argument is not a fair account of the field and experienced examiners will notice. Engaging with conflicting evidence, and explaining how your work relates to it, is a sign of intellectual honesty and sophistication.

Treating older sources as irrelevant. Some of the most important theoretical and methodological contributions in a field are decades old. Currency is not the same as importance. Know which are the foundational texts in your area and engage with them even if they're not recent.

No clear argument about what the review shows. A review that describes the field but doesn't draw a conclusion about the state of knowledge, or doesn't connect to the research question it's supposed to contextualise, has not done the work a literature review is supposed to do.

Using tools to manage the process

The volume of sources in a literature review benefits from systematic organisation. Fabric's semantic search means you can find relevant notes across your entire library by describing what you're looking for rather than remembering which folder you put things in. Annotations on PDFs make highlights and comments searchable and linkable to other notes. The AI assistant can synthesise across your collected sources: "What have I found that relates to the methodological debate about X?" saves significant time at the synthesis stage.

Reference management software (Zotero, Mendeley, EndNote) handles citation formatting and can import references directly from databases. Keeping your PDF library and your reference manager in sync prevents the perennial problem of notes that reference something you can't find.

Frequently asked questions

How long should a literature review be?

It depends on the context. A dissertation literature review might be 5,000 to 15,000 words depending on the level and the field. A journal article literature review might be 1,000 to 3,000 words. Standalone systematic reviews have their own conventions. Your supervisor or the submission guidelines will specify the expected length for your context.

How many sources do I need?

Enough to demonstrate comprehensive coverage of the relevant field within your defined scope, and enough to support your argument about the state of knowledge. For a doctoral dissertation this might be 100-200 sources; for an undergraduate dissertation, 30-60 might be appropriate. Ask your supervisor what's expected in your field.

How do I avoid plagiarism in a literature review?

By summarising and paraphrasing in your own words rather than copying phrases, citing every source you draw on (including for ideas and arguments, not just quotes), and being clear about which ideas are yours and which come from sources. Putting citations at the end of long paragraphs that contain many source ideas is not adequate; cite closely and specifically.

What's the difference between a narrative review and a systematic review?

A narrative review draws on sources selected by the author's judgement of relevance and quality. A systematic review uses a documented, reproducible search strategy and explicit inclusion/exclusion criteria, and often uses formal quality appraisal tools. Systematic reviews are designed to minimise bias and are standard in healthcare and policy research. Most dissertation literature reviews are narrative reviews with some systematic elements.

How do I handle sources I can't access?

For journal articles, check whether your institution has access through its library portal, use interlibrary loan, email the corresponding author (who will almost always send you a copy), or use open access repositories like PubMed Central, arXiv, or SSRN. If a source looks central to your topic and you really can't access it, note this limitation in your methodology section.

When should I stop searching and start writing?

When new searches are returning sources you've already found or sources that aren't significantly changing your understanding of the field. When you can characterise the main positions, debates, and gaps without needing to look things up. You won't have read everything; the question is whether you've read enough to make a defensible account of the relevant literature.