How to Read a Scientific Paper — A Researcher's Guide

Pass 1 — The 10-minute skim (decide if it's worth reading)

The first pass answers one question: should you read this paper at all? Spend 10 minutes, no more. Read in this order: (1) title and abstract, (2) introduction's last paragraph — which usually states the hypothesis and summarises the findings, (3) figure legends for every main figure, (4) conclusion or discussion's first paragraph. You are looking for three things: the research question, the main claim, and the type of evidence.

Do not read the methods, do not read the results prose, do not read the discussion's nuance. At the end of 10 minutes write two sentences in your notes — what the paper claims, and how they tested it. If you cannot write those sentences, the paper is either badly written or not worth your time. This pass is also where AI summarisers earn their place: a two-paragraph AI summary is a reasonable substitute for the first-pass skim on papers you are triaging, not papers you are citing.

Pass 2 — The 45-minute read (understand the paper)

If pass 1 clears the bar, commit 45 minutes for pass 2. Read in a different order from the paper's structure: (1) introduction, fully, to understand what question the authors are trying to answer and what prior work they build on; (2) all figures and tables, in order, with their legends — do this before reading the results prose; (3) the results section, using the figures you have already seen as scaffolding; (4) the discussion. Skim the methods for technique names but do not read them in detail yet.

By the end of pass 2 you should be able to state, in your own words, the experimental design, the primary result, the authors' interpretation, and at least one limitation they acknowledged. If you are taking notes for a literature review, this is the pass where you populate the extraction fields: sample size, population, intervention, comparator, outcome, effect size, limitations. See our guide on how to summarise a research paper for a structured template.

Pass 3 — The deep read (reproduce the paper mentally)

Pass 3 is for papers that are load-bearing for your own work — papers you will cite as evidence for a claim, or whose method you plan to use. Budget 2–4 hours. Now you read the methods in full, well enough that you could (in principle) replicate the experiment. Check specifics: antibody catalogue numbers, cell line sources and passage numbers, statistical tests used and whether they fit the data, sample size justification, blinding and randomisation, exclusion criteria.

Check the supplementary materials — load-bearing controls and negative results usually live there. Read the discussion critically: are the conclusions actually supported by the figures, or are they over-reaching? If the paper cites a prior result as foundational, pull that reference too. This is also where you check for retractions (PubMed's retraction notice, Retraction Watch) and for critical comments on PubPeer. The three-pass model is sharper on biomedical papers than on theory-heavy fields, but the principle holds: read in layers, not linearly.

What to skim, what to scrutinise

Different sections warrant different attention depending on your goal. Skim: the introduction's first half (context-setting prose), the discussion's speculative later paragraphs, and author-bio sentences. Scrutinise: sample size and power calculations, the exact wording of the primary outcome (prespecified? or defined post hoc?), which statistical tests were used and whether assumptions are met, the control conditions in every figure, error bars (SD vs SEM vs 95% CI — these are not interchangeable), and the "n =" values on every figure — especially whether n refers to biological replicates, technical replicates, or animals.

Also scrutinise the funding statement and competing interest disclosures; they do not automatically invalidate a paper but they contextualise it. In clinical research papers, scrutinise inclusion/exclusion criteria and CONSORT-style flow diagrams. If you are reading a meta-analysis, scrutinise the search strategy, the PRISMA diagram, and the heterogeneity statistics (I²).

How to read figures before prose

Papers are written prose-first, but good readers read figures-first. This is because figures carry the actual evidence and the prose is the author's interpretation of it. Workflow: open the PDF, jump to Figure 1, read the legend carefully, then examine every panel. For each panel ask: what is plotted, what are the axes, what comparison is being shown, what is the sample size, and what is the direction and magnitude of the effect?

Do this for every main figure before reading the results prose. When you then read the prose you will catch over-statements: "significantly reduced" when the effect is 8% with n = 3, "robust" when error bars span the control mean, "dose-dependent" when only two doses were tested. For journals that place figures at the end of the PDF, keep two windows open. For journals with inline figures (Nature, Cell, Science), you can read down the PDF directly. Figure-first reading is the single biggest upgrade most researchers can make to how they read research papers.

When to read full text vs when to rely on abstract

A good abstract captures the question, method, result, and conclusion in 250 words. For most screening decisions the abstract is enough. Read the full text when: (1) you plan to cite the paper as evidence for a specific claim — the abstract may omit the exact effect size, (2) you plan to use the paper's method, (3) the abstract's claim seems surprising relative to the field, (4) you are writing a systematic review or meta-analysis, or (5) the paper is load-bearing for your own hypothesis.

Relying on abstracts alone is how error propagates through literature: a nuanced result ("effect observed in subgroup A only, with caveats") becomes "effect observed" in the citing paper's introduction, then "established" three citation hops later. Do not be the citation that loses the caveat. For a triage-heavy workflow, AI research paper summarisers can expand abstract content usefully — but they can also smooth over caveats, so verify any load-bearing claim in the full text.

Common mistakes to avoid

• Reading linearly from start to finish. Papers are not novels. Skip around; abstract → figures → results → intro → discussion → methods is a better order.
• Trusting the abstract for a citation. Abstracts flatten nuance. If you are going to cite it, read the paper.
• Skipping the supplementary. In biomedical papers, the supplementary often contains the controls and negative results that qualify the main-text claims.
• Not checking sample size or n. A three-panel figure with n = 3 per group is a pilot, not evidence. Always check.
• Ignoring retractions and corrections. Check PubMed for retraction notices and search Retraction Watch and PubPeer for critical comments before citing a paper for the first time.
• Letting an AI summariser replace reading. AI summaries are a triage tool, not a reading substitute. For any paper you cite, read the figures and the methods yourself.

Tools and resources

• BioSkepsis — biomedical AI research assistant; can summarise, extract methods details, and cross-check claims against 40M+ peer-reviewed papers. Free tier (100 papers/session).
• Scite — shows how every citation has been used (supporting, contrasting, mentioning); useful when evaluating a paper's reception.
• PubPeer — post-publication peer review; check here for flagged concerns on a paper.
• Retraction Watch — database of retractions and expressions of concern.
• Zotero — free reference manager with PDF annotation; use it to keep structured notes while you read.

For a wider perspective on AI-assisted reading, see our ranked guide to AI research tools.

How BioSkepsis helps with this

BioSkepsis accelerates passes 1 and 2 without replacing pass 3. Paste a DOI and get a grounded summary of the question, method, and primary result, with every claim linked to the source paragraph — so you can jump straight to the figures you need to verify. For methodology comparison across related papers, the mechanistic-links table surfaces how different studies converge or diverge on the same claim. If you are reading a paper whose field you are new to, BioSkepsis can pull the five papers it most depends on and explain them in context — useful when a paper assumes background you do not have. See the AI research paper summariser for the specific workflow.

Frequently asked questions