How to write a literature review: a step-by-step guide

A literature review shows what is already known about a question, how that knowledge fits together, and where it runs out. Done well, it is the foundation of a thesis, a grant proposal or a paper. Done badly, it reads like an annotated list of everything you happened to read. This guide walks through the whole process, from choosing a scope to revising the final draft.

What a literature review is for

A literature review is not a summary of papers. It is an argument about a body of research. It should tell the reader what the main findings are, where studies agree and disagree, which methods the field relies on, and what has not been answered yet. That last part, the gap, is usually why the review exists: it sets up the question your own work will address.

Reviews come in several forms. A narrative review, the kind most theses and papers include, surveys the literature selectively to build an argument. A systematic review follows a predefined protocol to find and appraise every eligible study. If you are not sure which one you need, read Systematic review vs literature review first.

Step 1: Define your question and scope

Start by writing your research question in one sentence. Then decide what is in and out of scope: the population or subject, the time period, the types of study and the disciplines you will cover. A review of "sleep and memory" has no edges; a review of "the effect of sleep deprivation on working memory in adults, in experimental studies since 2000" is something you can finish.

Write these boundaries down. You will come back to them every time you wonder whether a paper belongs, and they will become part of your introduction.

Step 2: Search the literature deliberately

Even a narrative review benefits from a planned search. Choose the databases that cover your field: PubMed for biomedicine, PsycINFO for psychology, ERIC for education, Web of Science and Scopus across disciplines, and Google Scholar for breadth. Build a search from your key concepts, their synonyms and related terms, combined with AND and OR.

• Keep a log of each search: the database, the exact search string, the date and the number of results.
• Use citation chasing: check the reference lists of key papers, and the newer papers that cite them.
• Set up alerts for your main searches so new papers reach you while you write.

Save everything you find to one place as you go, with its DOI or PubMed ID, so you never search for the same paper twice. How to organize research papers describes a system that holds up past a few hundred sources.

Step 3: Screen and select

Not everything you find belongs in the review. Screen in two passes: first titles and abstracts against your scope, then the full text of the papers that pass. Note briefly why you excluded each full text. If your choices are ever questioned, or you need to update the review next year, that record saves you from redoing the work.

Step 4: Read actively and take notes that synthesize

Read with your question in mind, not the paper's. For each study, capture the same core details so you can compare them later:

• the research question and design;
• the sample or data, and its size;
• the main measures and methods;
• the key findings, with numbers where they matter;
• limitations, including ones the authors do not mention.

A synthesis matrix makes this concrete: one row per study, one column per detail. Patterns that stay invisible across thirty PDFs become obvious in a table. If reading itself is the bottleneck, How to read a scientific paper covers a faster method.

Step 5: Find the structure

Once your notes sit in one place, look for the shape of the field. Most reviews are organized in one of four ways:

• Thematically, around the main ideas or debates. This is the most common choice and usually the strongest.
• Methodologically, comparing what different approaches have found.
• Chronologically, tracing how understanding developed over time.
• Theoretically, comparing the frameworks used to explain the evidence.

Write an outline with a heading for each section and a one-line claim under each heading. If you cannot state the claim yet, you have a topic, not a section, and more reading or thinking is needed before you draft it.

Step 6: Write the draft

A literature review has the same three parts as any argument:

• An introduction that states the question, why it matters and the scope you chose.
• A body that works through your themes, comparing and evaluating studies rather than describing them one at a time.
• A conclusion that summarizes what is known, names the gaps and shows how they lead to your own question.

The most common weakness in a review is summary without synthesis: paragraph after paragraph, each describing a single study. Instead, open each paragraph with a point and use several studies as evidence for it. "Three experimental studies found an effect, but all three measured it within a day, and the one longitudinal study found it faded after six months" does more work than three separate summaries.

Be explicit about quality as well as quantity. Ten small, uncontrolled studies pointing one way can matter less than one large, well-designed trial pointing the other. Your reader needs your judgment, not only your list.

Step 7: Cite carefully and revise

Cite every claim that comes from a source, and check that the source really says what you attribute to it. Use one citation style consistently: the one your department or target journal requires. If you are unsure which applies, see APA vs MLA vs Chicago.

Then revise in three passes. Read once for argument: does each section make a claim and support it? Read once for coverage: is any important study missing? Read once for citations: does every reference appear in the list, and does every listed item appear in the text?

A short checklist

• The question and scope are stated in the introduction.
• The search is documented well enough that someone else could repeat it.
• Each section makes a claim, supported by several studies.
• Disagreements between studies are explained, not just reported.
• The conclusion names the gap your work addresses.
• Every citation is accurate and in one consistent style.

Doing it in Mosaic

Mosaic keeps a review in one project: the papers, your highlights and notes, the tasks and the drafts. You can import from Zotero or Mendeley, add papers by DOI or PDF, ask the assistant to compare studies with citations to your library, and cite as you write in Google Docs or Word. A literature review, start to finish, in Mosaic walks through each step.