How to Read a Scientific Paper: A Step-by-Step Method

You open the PDF. You start at the top, because that is how reading works. Three paragraphs into the introduction you hit a sentence with four acronyms and two citations you have never heard of, and by the end of the page, you are rereading the same lines without taking anything in. Forty minutes later you have highlighted half the paper and could not say what its main argument is. You worry, rightly so, that you are missing important information. But there’s only so much time in the day to figure it out.

If that is familiar, the problem is not you. It is the order you are reading in. Research articles were never meant to be read start-to-end liek novel. This guide gives you a different order–a hunting order–five questions to ask of any research article, and a way to know when you are done. It works for undergraduates reading their first primary article, graduate students facing a long reading list, and anyone outside academia who needs to understand a study, for example if a loved becomes seriously ill, rather than skim a news story about it.

Why reading start to finish fails

Research articles are written for other experts, and they are organized to satisfy journals, not to help readers identify the main argument. The standard structure (introduction, methods, results, and discussion, aka IMRaD) front-loads exactly the material a newcomer is least equipped to handle: background, jargon, and a review of prior work whose point you cannot see yet, because you do not know what the study found.

Experienced scientists almost never read an article in this order. They go looking for the main finding, check how it was supported and developed, and only then read the background to see where it fits. The method below makes that expert habit explicit so you can practice it deliberately.

Before you start: five minutes of orientation

Read the title, the abstract, and the captions of every figure and table. Do not try to understand the figures yet; just note what kind of data the paper contains. Then write down, in one line, what you think this paper is about. that is the topic. In the next step, you will see that the topic and the claim are different things, and differentiating them is the first real skill of critical reading.

The five questions

Every research article, in every scientific field, is built from the same five elements: a Claim, the Evidence for it, the Reasoning that connects the two, the Implications of the result, and the Context that motivated the study. Together they form CERIC representing the categories of information that we hunt for in that order. If you’re new to the topic, read for the Claim first, then for what supports it, and you save the background for last. If you’re very experienced in the topic, you might change up the order, and that’s totally fine. The point is to do what works best for you to develop your critical reading skills. Afterall, CERIC is a hunting map, not gospel.

1. Claim: what new knowledge does this article assert?

The Claim is the article’s central conclusion, the new thing the authors say they now know. It is not the topic (“gut bacteria and anxiety”) but a statement about the topic (“transplanting gut bacteria from anxious mice into calm mice increased anxiety-like behavior in the recipients”).

Where to look: the title, the last sentences of the abstract, and the first paragraph of the discussion or conclusion. Compare what is said in different locations. Authors sometimes state the Claim more boldly in one place than the other, and the difference tells you something.

Write the Claim in your own words, in one sentence. If you cannot, keep hunting. Here is a simple test worth applying: a genuine Claim is supported by Evidence and Reasoning inside this study. A statement supported only by citations to other studies is Context, and a statement the authors hope for the future but did not test is an Implication. Many articles make more than one Claim (called subclaims or secondary claims). Focus the main one first.

2. Evidence: what did they actually observe or measure?

Now go to the figures, tables, and results section, and find the specific data that support the Claim you wrote down. Not all of the data; only the data for that Claim. Usually one or two figures carry the main result and the rest are controls, checks, or secondary findings.

For each piece of evidence, write what it shows in plain language: what was compared with what, how big the difference was, and how much uncertainty the authors report. Then glance at the methods to see how that evidence was produced. You do not need to understand every procedure; you need to know what was measured, in whom or what, and how many times.

Our companion guide, How to Evaluate the Evidence in a Research Article, goes deeper on judging whether the evidence is good enough.

3. Reasoning: how do the authors get from the evidence to the claim?

Evidence does not speak for itself. Between “the recipient mice spent less time in the open arms of the maze” and “gut bacteria cause anxiety-like behavior” sits a chain of logical interpretation: that the maze measures anxiety, that the transplant was the only relevant difference between the groups, that the effect is not an artifact of the procedure. That chain is the Reasoning, and it is where most weak papers are weak or break.

Where to look: the discussion section, and the sentences in the results that begin with words like “suggesting,” “indicating,” or “consistent with.” It’s everywhere and nowhere. So, ask three things: What assumptions does the argument rely on? What alternative explanations did the authors rule out, and how? What alternatives did they not mention? A good article addresses its own limitations; a great one tells you what evidence would overturn its Claim.

4. Implications: why does this matter, and what comes next?

Implications are the future tense of the article: what the authors think should happen now that the Claim is on the table. Some are narrow (the next experiment), some are broad (changes to clinical practice, policy, or theory). Look in the final paragraphs of the discussion and in the abstract’s closing sentence.

Two things to check. First, are the stated Implications proportionate to the Claim? A study in forty mice can suggest a direction for human research; it cannot recommend a treatment. Second, are there Implications the authors did not state? Often the most interesting consequence of a finding is one the authors, focused on their own field, did not think to mention.

5. Context: what gap was this study designed to fill?

Only now do you read the introduction. You will find it far easier than it looked an hour ago, because you already know what the study found and you are reading the background for one specific purpose: to find the rationale. Somewhere in the introduction, usually near its end, the authors say what prior work established, what remained unknown, and why that unknown mattered enough to study. That gap is the Context.

Context is the past tense of the article, the counterpart to Implications. Reading it last lets you judge it: does the Claim actually fill the gap the authors describe, or did the study answer a slightly different question than the one it set out to answer? That mismatch is common and easy to miss when you read in order.

Finishing: the five-sentence review

You are done when you can write five sentences, one per element, in your own words. Claim, Evidence, Reasoning, Implications, Context. That short review is more useful than any amount of highlighting. It is what you will consult when you cite the paper in six months, what you can bring to a journal club or seminar, and the raw material for a literature review.

If one of the five sentences is hard to write, that difficulty is information. An article whose Reasoning you cannot summarize may not have made its Reasoning explicit. An article with no identifiable Implications may be a solid but minor result. The method does not just help you understand papers; it helps you notice where they are thin.

Common mistakes

  • Confusing the topic with the Claim. “This paper is about X” is not a Claim. “This paper shows that X does Y under condition Z” is.
  • Treating every citation as evidence. Evidence is what this study produced. What other studies found is Context.
  • Highlighting instead of writing. Highlighting feels like progress and produces nothing you can use later. One sentence per element, in your own words, does.
  • Starting with the methods. Methods only make sense once you know what evidence they were meant to produce.
  • Giving up on the first pass. Your first article with this method may take an hour or more. That is normal. The time drops steeply with practice because the five questions are the same every time, in every field.

Keep the five questions beside you

The free CERIC cheat sheet puts the five elements, where to find each one in a typical article, and the question to ask on a single printable page. Keep it next to the next paper you read.

If you want to practice with guidance, Foundations Course 1 is free and walks through the method on real articles. The full treatment, with worked examples from across the sciences and chapters on using the method for critique, literature reviews, and journal clubs, is in How to Critically Read the Scientific Research Literature (Cambridge University Press, 2026). Instructors can find a free teaching pack on the For Educators page.