Counselor intelligence brief

Regeneron Science Talent Search

A practical planning guide for turning serious high-school research into a Regeneron STS submission: eligibility, strategy, timeline, and the kind of project architecture that can credibly compete.

Build the strategy Study winners Check readiness

Snapshot date May 21, 2026. Society for Science says the 2027 application opens June 1, 2026 and closes in early November 2026.

1. What Regeneron STS is

Regeneron STS is the United States' oldest and most prestigious science and math competition for high-school seniors. It is run by Society for Science and sponsored by Regeneron, with 300 scholars, 40 finalists, and more than $1.8 million in awards each cycle.

The core product is not a poster. It is a full application built around an original, independent research report, essays, recommendations, transcripts, and evidence that the student has unusual scientific promise. Treat it like a research thesis plus a leadership file.

Applicants

2,600+

2026 was the largest applicant pool since 1967.

Scholars

300

Each scholar and school received a $2,000 award in 2026.

Finalists

Finalists compete in March for top national awards.

2. Eligibility and cycle timing

The first screen is procedural. If the student or project fails here, stronger science will not rescue the application.

A strong STS candidate usually has

Original research conducted during high school.
A clear individual contribution that can be defended without a teammate.
A mentor who advises without becoming a hidden co-author.
Enough time to write, revise, document approvals, and collect recommendations.

The common disqualifiers

Team research presented as individual work.
Missing pre-approval or documentation for regulated research areas.
Student is not in the eligible senior-year category.
Late application, late recommendation, or unresolved application issue after the support window closes.

2027 cycle planning dates

Application opens

June 1, 2026
Application closes

Early November 2026
Customer support cutoff

24 hours before deadline
Scholars named

January 2027 expected
Finalists named

Late January 2027 expected
Finalist week

March 2027 expected

3. The project that can compete

Regeneron rewards more than technical polish. The entry needs a scientific claim, a believable method, and visible student judgment.

What judges are scoring

Originality

A real question, not a packaged demonstration or polished class project.
Scientific rigor

A research report that can survive Ph.D.-level review, including limitations.
Independence

The idea, execution, analysis, and writing must be the student's own work.
Leadership promise

Essays and recommendations need to show the student behind the science.

How to shape the research

Anchor the question

Start with a mechanism or phenomenon the student found, then make the model serve that discovery.
Beat a baseline

STS readers need to know what changed because of the student's work. A credible benchmark is cleaner than a dramatic adjective.
Show the student's judgment

A strong entry explains why tempting alternatives were rejected, not just what finally worked.
Keep the paper defensible

Every figure should connect to the central claim, with no cosmetic charts padding the report.

Client strategy pattern

Discovery first, model second

For an Avery-style climate project, the clean STS arc is: build or discover a novel regime, show why existing models miss it, then design a predictive model that uses the discovery and beats a published baseline in the regime where the old approach fails.

1. Novel artifact

Catalogue, dataset, proof, sensor, assay, or pipeline.

2. Mechanism

What the artifact reveals that was not visible before.

3. Benchmark

A fair baseline that makes improvement measurable.

4. Interpretation

Why the result happened and when it does not generalize.

4. Winners to study

Recent winners are useful because they show the range of projects STS will reward. The lesson is not to copy a field, but to study the shape of the contribution.

2026 top three

Noble polyhedra, retina-based AI screening, and clam cancer models

View winners

1st place

Connor Hill

Completed a computational and proof-driven classification of noble polyhedra, expanding the known set of highly symmetric shapes.
2nd place

Edward Kang

Built RetinaMind, an AI and cell-model approach to screening for neurodevelopmental disorders from retinal signals.
3rd place

Iris Shen

Tested whether clams with naturally occurring blood cancer can serve as a practical model for leukemia drug discovery.

Counselor read: The top three are wildly different fields, but each has a crisp artifact: a complete classification, a working screening model, or a new organism model.

2025 top three

Infrared astronomy, rare-disease treatment, and hypergraph math

View winners

1st place

Matteo Paz

Used waveform-based machine learning on nearly 200 terabytes of WISE telescope data to identify infrared variable objects.
2nd place

Ava Grace Cummings

Created a fruit-fly model of STAC3 disorder and tested candidate treatments tied to a rare muscular condition.
3rd place

Owen Jianwen Zhang

Solved a theoretical combinatorics problem involving 3-uniform hypergraphs.

Counselor read: A winning STS project can be computational, experimental, or theoretical. The constant is original work with a concrete contribution.

2024 top three

AI interpretability, cancer metabolism, and mathematical optimization

View winners

1st place

Achyuta Rajaram

Improved automatic discovery of visual circuits inside machine-learning models, making image classifiers more interpretable.
2nd place

Thomas Cong

Studied how metabolism and gene expression interact in immune cancers, challenging a simple gene-expression-only view.
3rd place

Michelle Wei

Advanced second-order cone programming solvers, a mathematical optimization tool used in many applied systems.

Counselor read: Top projects often make a hard system more legible: an AI model, a cancer pathway, or a solver that underlies real optimization problems.

5. Roadmap to submission

Work backward from early November. The submission month is for writing and quality control, not discovering whether the experiment works.

May to June

Lock mentor boundaries, confirm individual ownership, audit prior research, and define the claim that would be worth a 20-page report.
June to July

Rebuild data and baselines, create reproducible notebooks, and write a short methods memo before adding complexity.
July to August

Run the main experiment, ablations, and negative controls. Freeze a versioned analysis pipeline.
September

Turn results into figures, compare against credible baselines, and collect enough failure cases to make the discussion honest.
October

Draft the research report, essays, and recommendation packet. Submit support tickets before the 24-hour support guarantee window.
Early November

Submit the full application and recommendations before the 8pm ET deadline once the 2027 exact date is posted.

6. Readiness check

Use this as the first advising screen before committing a student to the STS sprint.

Eligible senior

Last year of secondary school in the U.S. or territories, or a U.S. citizen abroad.
Individual research

No team project split apart for STS, even if roles were different.
Original report

Maximum 20-page scientific paper, written by the student, with appropriate citations and documentation.
Documentation clean

Human subjects, animals, tissues, cell lines, pathogens, and hazardous agents need rules review before submission.
Recommendation plan

Teachers and mentors need deadlines, project context, and enough lead time to write specifically.

If time is tight

Cut scope before cutting rigor. A smaller claim with clean evidence beats a sprawling project with soft controls.

If the mentor is strong

Define advisory boundaries in writing. STS wants independence, not invisible lab authorship.

If the result is negative

A negative result can still work if the question mattered, the test was fair, and the explanation is scientifically useful.

7. Official resources

Keep these links close. The 2027 official rules are expected in June 2026, so the site should be refreshed after Society for Science posts the new packet.