Last Updated on July 16, 2026 by Dr. Bhagat
What is the most cited research paper in the world? Which papers are shaping scholarship right now? In this guide, we unpack the most cited research papers in 2025, why they attract so many citations, and how you can apply those lessons to grow your own research impact.
Quick answer: What is the most cited paper in the world?
“Protein measurement with the Folin phenol reagent” (Lowry et al., 1951) is widely recognized as the most cited paper of all time in citation databases, thanks to its foundational lab protocol used across biology and biochemistry.
Why people search this topic (and what you’ll get here)
- A concise list of highly cited research papers (timeless classics + 2025-relevant standouts).
- Actionable lessons you can use to increase citations and visibility.
- Clear definitions of citation metrics and how they differ.
Average patterns among the most cited research papers (2025)
Across fields, the most cited papers tend to be:
- Methods papers that become standard protocols (e.g., Lowry protein assay).
- Tools/frameworks that enable downstream research (e.g., statistical methods, sequencing pipelines, deep-learning architectures).
- Large, collaborative datasets and consortia with broad reuse.
- Open access or easy-to-reuse resources with permissive licensing.
Top highly cited research papers: timeless anchors
- Lowry et al. (1951) — Protein measurement with the Folin phenol reagent. A core biochemical method used for decades.
- Laemmli (1970) — Cleavage of structural proteins during SDS-PAGE. The backbone of modern protein electrophoresis.
- Bradford (1976) — Rapid protein quantitation. A fast alternative colorimetric assay, widely adopted.
- Chomczynski & Sacchi (1987) — Single-step RNA isolation. A simple guanidinium-thiocyanate–phenol–chloroform method used in molecular labs worldwide.
- Schneider et al. (2012) — ImageJ 25 years. A landmark tool paper documenting a ubiquitous image analysis platform.
Most cited research paper in 21st Century
Google Scholar [Provided March 2025]
| Rank | Title | Year | Cited by |
| 1 | Diagnostic and Statistical Manual of Mental Disorders, DSM-5 | 2013 | 367800 |
| 2 | Deep Residual Learning for Image Recognition | 2016 | 254074 |
| 3 | Using thematic analysis in psychology | 2006 | 230391 |
| 4 | Adam: A Method for Stochastic Optimization | 2014 | 201831 |
| 5 | Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2−ΔΔCT Method (2001) | 2006 | 185480 |
| 6 | Attention is all you need | 2017 | 150832 |
| 7 | Random forests | 2001 | 146508 |
| 8 | ImageNet classification with deep convolutional neural networks | 2012 | 137997 |
| 9 | Very Deep Convolutional Networks for Large-Scale Image Recognition | 2014 | 137214 |
| 10 | Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement | 2009 | 134519 |
| 11 | BERT: Pre-training of deep bidirectional transformers for language understanding | 2019 | 125595 |
| 12 | Scikit-learn: Machine learning in Python | 2011 | 105225 |
| 13 | U-Net: Convolutional Networks for Biomedical Image Segmentation | 2015 | 100673 |
| 14 | A short history of SHELX | 2007 | 99470 |
| 15 | Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries | 2021 | 99390 |
| 16 | Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries | 2018 | 93433 |
| 17 | Deep learning | 2015 | 90674 |
| 18 | Fitting Linear Mixed-Effects Models Using lme4 | 2015 | 86931 |
| 19 | Common method biases in behavioral research: A critical review of the literature and recommended remedies. | 2003 | 84589 |
| 20 | Cancer statistics | 2018 | 80338 |
| 21 | Hallmarks of Cancer: The Next Generation | 2011 | 80093 |
| 22 | ImageNet: A large-scale hierarchical image database | 2009 | 79921 |
| 23 | The PRISMA 2020 statement: an updated guideline for reporting systematic reviews | 2021 | 79476 |
| 24 | Electric Field Effect in Atomically Thin Carbon Films | 2004 | 79165 |
| 25 | Generative adversarial nets | 2014 | 77764 |
| 26 | Distinctive Image Features from Scale-Invariant Keypoints | 2004 | 77567 |
| 27 | Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 | 2014 | 76019 |
| 28 | G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences | 2007 | 73578 |
| 29 | Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being | 2000 | 71884 |
| 30 | Going deeper with convolutions | 2015 | 64739 |
| 31 | Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China | 2020 | 64091 |
| 32 | NIH Image to ImageJ: 25 years of image analysis | 2012 | 63516 |
| 33 | Fiji: an open-source platform for biological-image analysis | 2012 | 61640 |
| 34 | Accelerating deep network training by reducing internal covariate shift | 2015 | 60166 |
| 35 | User acceptance of information technology: Toward a unified view | 2003 | 58382 |
| 36 | Image quality assessment: from error visibility to structural similarity | 2004 | 58281 |
| 37 | You only look once: Unified, real-time object detection | 2016 | 57286 |
| 38 | Measuring inconsistency in meta-analyses | 2003 | 57246 |
| 39 | Trimmomatic: a flexible trimmer for Illumina sequence data | 2014 | 56719 |
| 40 | Three approaches to qualitative content analysis | 2015 | 56431 |
| 41 | The sequence alignment/map format and SAMtools | 2009 | 56294 |
| 42 | Latent dirichlet allocation | 2003 | 55893 |
| 43 | Dropout: a simple way to prevent neural networks from overfitting | 2014 | 55010 |
| 44 | Microsoft COCO: Common objects in context | 2014 | 54421 |
| 45 | Fully convolutional networks for semantic segmentation | 2015 | 54265 |
| 46 | The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration | 2009 | 53821 |
| 47 | An image is worth 16×16 words: Transformers for image recognition at scale | 2021 | 53705 |
| 48 | A fast and elitist multiobjective genetic algorithm | 2002 | 53556 |
| 49 | Pytorch: An imperative style, high-performance deep learning library | 2019 | 52586 |
| 50 | Fast gapped-read alignment with Bowtie 2 | 2012 | 52071 |
Highly cited in recent years (relevant in 2025)
In the last decade, several papers have amassed fast-rising citation counts due to transformative impact:
- He et al. (2016) — Deep Residual Learning for Image Recognition (ResNet). A foundational AI architecture used across computer vision.
- Vaswani et al. (2017) — Attention Is All You Need. Introduced the Transformer, the basis for today’s large language models.
- Jinek et al. (2012) / Doudna & Charpentier team — Programmable genome editing with CRISPR-Cas9. Sparked an explosion of gene-editing research and applications.
- Large COVID-19 vaccine/epidemiology studies (2020–2022). High reuse across medicine and public health.
How these papers win citations (repeatable patterns)
- They solve a frequent, painful problem (e.g., quick protein quantitation; scalable model training).
- They make reuse easy (clear protocols, code, datasets, permissive licenses).
- They publish where their audience is (journals with strong reach and indexing).
- They keep improving (follow-up releases, better baselines, updated documentation).
Snippet corner: common questions answered fast
What is the most cited paper in the world?
Lowry et al. (1951) on protein measurement is widely regarded as the most cited paper of all time.
Which fields produce the most highly cited papers?
Technique-heavy life sciences, computer science (AI), and large-scale biomedical studies frequently generate the most citations due to broad reuse.
Examples table: influential papers and why they attract citations
| Paper (short) | Field | Why it’s highly cited |
|---|---|---|
| Lowry et al. (1951) | Biochemistry | Universal protein assay; cited whenever used. |
| Laemmli (1970) | Molecular Biology | Standard SDS-PAGE method; ubiquitous protocol citation. |
| He et al. (2016) — ResNet | AI / CV | Backbone for countless models; benchmarks and transfers. |
| Vaswani et al. (2017) — Transformer | AI / NLP | Core architecture behind modern LLMs; massive reuse. |
| Jinek et al. (2012) — CRISPR-Cas9 | Genomics | General-purpose genome editing; cross-disciplinary uptake. |
How to learn from the most cited research papers (and apply it)
1) Design for reuse
- Publish clear, reproducible protocols or well-documented code.
- Choose repositories with stable DOIs and long-term preservation.
2) Target the right venue
- Match your audience’s reading habits and indexing coverage.
- Compare options via what is impact factor and top journals 2025.
3) Make the first citation easy
- Provide a copy-paste Cite this work snippet in BibTeX and APA.
- Include minimal working examples and starter notebooks.
4) Share early, then iterate
- Preprints, code, and datasets can accelerate feedback and adoption.
- Ship improvements (v1.1, v2) and changelogs so scholars can track progress.
Key metrics (and what they really mean)
- Total citations: raw count of times a work is cited.
- Field-normalized impact: adjusts for different citation cultures.
- Journal impact factor: a journal-level average; use with caution. See our explainer: what is impact factor.
- Altmetrics: attention beyond citations (news, policy, social).
FAQ
How do citations impact a research paper’s success?
Citations signal that other scholars have found your work useful. They can influence hiring, promotion, funding, and collaborations, though context matters by field.
What’s the fastest way to get more citations?
Make reuse easy: release code/data, write a short “Getting started” section, and choose venues with the right audience. Then share updates and respond to issues.
Which fields had the most cited papers in 2025?
Life-science methods, AI architectures, and large-scale biomedical studies continued to dominate because they enable broad reuse across disciplines.
Is journal impact factor the best predictor of citations?
Not always. Impact factor is a journal average. Paper-level quality, relevance, and accessibility often matter more for your own citation trajectory.
Conclusion
The most cited research papers — from Lowry’s assay to modern AI and CRISPR — win because they solve common problems and enable others. Design for reuse, share early, and keep improving. Which lesson will you apply to your next manuscript?
