How to Take Notes for a Literature Review (5 Steps)

Published October 11, 2026 · Updated October 11, 2026 · 8 min read · by the Sakura5 team

The best way to take notes for a literature review is to give every source the same short, fixed set of fields — its question, method, main finding, limitations, and how it relates to your own question — and then reorganize those notes by theme rather than by author before you write a word of the review. Notes organized source by source produce a review that reads like an annotated bibliography. Notes organized by theme produce an argument.

That distinction is the whole game. Supervisors and reviewers rarely complain that a literature review is missing a paper. They complain that it summarizes papers one after another without saying how they fit together, where they disagree, or what nobody has studied yet. Almost every time, that problem starts in the notes, long before the drafting.

This post covers why the usual summarize-each-paper habit breaks down, a five-step workflow for literature review notes, a worked example with 30 sources on a real-sized topic, how to handle the awkward cases, and a short FAQ.

Why summarizing each paper doesn’t add up to a review

The default approach goes like this: open a PDF, read it, write a paragraph summarizing it, move to the next one. After a month you have forty paragraphs in a document, each faithful to its paper, and no idea how to turn them into chapter two.

The trouble is that a literature review isn’t a list of what each author said. It’s a map of a conversation. As the UNC Writing Center’s guide to literature reviews puts it, the job is to synthesize and evaluate sources according to the guiding concept of your research, not simply to describe them. Free-form summaries make that hard in three ways:

• They aren’t comparable. One summary dwells on the method, the next on the results, the third on the theory. When you want to ask “which studies used interviews?” you have to reread all forty.
• They hide your opinion. A summary records what the author thinks. The review needs what you think about the author: whether the sample was too small, whether the finding contradicts someone else.
• They lock in the wrong order. Notes written in reading order push you to write in reading order, which is almost never the order of the argument.

How to take notes for a literature review: the workflow

The steps below work for a thesis chapter, a journal article’s background section, or a standalone review paper. The core idea is to separate two jobs that usually get mashed together: capturing each source consistently, and then synthesizing across sources.

1. Write your review question before you read

One or two sentences: what are you trying to find out from the literature? “What is known about how hybrid work affects early-career employees’ sense of belonging?” is a usable question. “Remote work” is a topic, and a topic gives you no filter for what to note. Your question decides which parts of each paper deserve a line and which deserve nothing.

2. Use the same template for every source

Before reading the first paper, fix the fields you’ll fill in for every one of them. A good starting set is: full citation, research question, method and sample, main findings, limitations, key terms or definitions, and a final field called “my take.” Keep each field to one or two lines. The point isn’t completeness — it’s that every note has the same shape, so you can compare them later.

3. Read for the template, not for everything

For most empirical papers you don’t need to read every page closely. Read the abstract, the introduction’s last paragraph (where the question usually lives), the methods, the results headings, and the discussion’s limitations section. Fill the template from memory first, then check back and add page numbers next to anything you might quote. If you want a deeper version of this read-then-recall habit, our guide to summarizing a long PDF into notes walks through it step by step.

4. Tag every note with themes

At the bottom of each note, add two to four short theme tags: “mentoring,” “office days,” “self-report only,” “contradicts Lee.” Don’t try to design the tag list up front. Let it grow for the first ten or so sources, then merge near-duplicates. These tags are what you’ll organize the review around, so they matter more than any single summary.

5. Rebuild the notes around themes

Once most of your sources are in, stop reading and reorganize. Put your review question in the center, each major theme as a branch, and the sources that speak to that theme underneath it, with a few words on what each one says. Then mark the tensions: two studies on the same branch that disagree, a branch with only one weak source, a question nobody has asked. Those marks become the “however” and “few studies have examined” sentences that make a review read like an argument.

A mind map is the most direct format for this step, because placing a source forces you to decide what it’s actually evidence for. In Mindmap AI, you can paste in the “findings” and “my take” fields from a batch of notes and get a first-draft map back in seconds (a Pro feature), then rename, merge, and drag branches until the structure reflects your reading rather than the AI’s guess.

A worked example: 30 sources for a master’s thesis chapter

Say you’re a master’s student in organizational psychology. Your thesis asks how hybrid work affects early-career employees’ sense of belonging, and your supervisor wants a draft literature review chapter in six weeks. A database search plus following references leaves you with 30 sources worth reading: 22 empirical studies, 5 review articles, and 3 theory papers.

Weeks 1–3: capture. You set up the seven-field template in a single document and work through about ten sources a week. Empirical papers take around 40 minutes each; the review articles take longer, and you use their reference lists to check you haven’t missed anything major. By week two your tag list has grown to 19 tags, so you spend twenty minutes merging it down to 8: belonging, mentoring, onboarding, office-day policy, manager behavior, measurement, cross-sectional only, and early-career specific.

Week 4: synthesize. You put the review question in the center of a map and make one branch per theme. Then you place every source. Three things jump out that no single summary showed you. First, the “mentoring” branch has eight sources, but only two of them looked at early-career employees specifically. Second, two studies on office-day policy reach opposite conclusions, and their “method” fields explain why: one surveyed employees at a single firm, the other compared several firms with different policies. Third, the “measurement” branch shows that belonging was measured with at least four different scales, which is a limitation worth a paragraph of its own.

Week 5: outline. The map turns almost directly into the chapter structure. Each theme branch becomes a section. Each tension becomes the turning point of that section. The thin early-career branch becomes your gap statement — the reason your study is needed. If you want a template for going from a map to an outline, our post on outlining from brainstorming notes applies just as well to a thesis chapter.

Week 6: draft. You write section by section, pulling citations and page numbers straight from the template notes. Because every note has the same fields, a sentence like “three of the five studies that used interviews found…” takes a minute to check instead of an afternoon.

What to do when the sources don’t fit the template

Not every source is a tidy empirical paper. A few adjustments keep the system working.

• Theory papers and books. Swap “method and sample” for “core argument” and “key concepts.” For a book, take notes per relevant chapter, not per book, and only on the chapters that answer your question.
• Review articles. Note their scope and date range, then use them as a checklist rather than a source of claims. Cite the original studies where you can, and record which ones you’ve actually read.
• Too many sources. If you have more than you can read in your timeline, fill only the citation, question, and findings fields for the weakest third, and give the full template to the studies your argument depends on.
• A theme that won’t settle. If a source keeps moving between branches, it probably belongs to both. Put it on one and add a cross-link to the other; that connection is often a sentence worth writing.

Keep a reference manager for citations and PDFs if you already use one. The template and the map do a different job: they hold your thinking about the sources, which a citation library doesn’t. Mindmap AI also has an outline view, which is a handy way to see the theme structure as a plain list once the map has settled.

FAQ

How many sources do I need for a literature review?

It depends on the field, the level, and how much has been published on your question, so ask your supervisor or check recent theses from your department. A better test than a number is saturation: when new sources keep landing on existing branches and stop changing your themes, you have enough to write.

Should I take literature review notes by hand or digitally?

Either works, but digital notes are easier to search, tag, and reorganize, which matters most for a literature review because you’ll rearrange notes by theme. If you read on paper, still transfer each source into the same digital template so you can compare sources side by side later.

What is a synthesis matrix?

A synthesis matrix is a table with sources as rows and themes or questions as columns, where each cell records what that source says about that theme. It does the same job as the theme map in this post. Use a matrix if you think in tables and a mind map if you think in relationships.

When should I stop taking notes and start writing?

Start drafting once your theme structure has stopped changing, even if a few sources are still unread. Writing exposes gaps faster than more reading does, and you can slot late sources into existing sections. Waiting until every paper is noted usually means waiting too long.

← All articles