Past science fair projects
What to do with 38,465 projects that have already been entered. The useful move is not to browse them for an idea — it is to treat them as prior work: find out what has been tried, find where somebody stopped, and start there.
1. Find out whether your idea has been done
Search the idea, not its keywords. A phrase like "detecting microplastics with a phone camera" will match projects that never used the word microplastic in their title. If a word-match search returns nothing, that means no project used those words — which is weak evidence about the idea itself, and is exactly when to search by meaning instead.
For a full proposal, the novelty check does this over the whole corpus at once and reports how crowded the neighbourhood is rather than just the top few matches. Check an idea →
2. Read the future-work paragraph, not the results
Most abstracts end with a sentence about what the student could not do: a control they lacked, a sample size that was too small, a condition they never tested, a material they wanted to try next. That sentence is the single most reliable source of a real project, because it is a gap identified by somebody who had already done the work to find it — and almost nobody reads it.
An idea taken from there is not a copy. You are picking up a thread the author put down and said so.
3. Use a past project without copying it
The line is the same one that applies in any field. Reproducing someone's method on a new population, material or condition is research. Reproducing their method on their material to get their result is not. Extending a result, contradicting it, or testing whether it holds outside the conditions it was measured in — all research.
Cite the project the way you would cite a paper: student, title, fair and year. Every project page here links back to the fair's own listing, which is the citable record; Fairfolio is where you found it, not the source. If your work started from someone's future-work line, say so in your own write-up. Judges read that as command of the field, not as weakness.
4. Calibrate scope before you commit
Read ten projects in one category and one year, placing and non-placing together. That cohort tells you what a finished project in that field looks like at your level far better than any individual winner does — how many trials is normal, what equipment is assumed, how narrow the question has to be to be answerable in a season.
- Physics and Astronomy 3,401
- Biomedical and Health Sciences 3,049
- Behavioural and Social Sciences 2,271
- Chemistry 2,217
- Animal Sciences 2,139
- Environmental Engineering 1,859
- Microbiology 1,825
- Cellular and Molecular Biology 1,751
- Plant Sciences 1,745
- Earth and Environmental Sciences 1,677
- Engineering 1,624
- Materials Science and Engineering 1,418
- Systems Software and Computer Science 1,007
- Biomedical Engineering 1,002
- Energy and Transport 878
- Robotics and Intelligent Machines 788
- Computational Biology and Bioinformatics 785
- Mathematics 769
- Computational Sciences 734
- Environmental Analysis 717
- Embedded Systems 522
- Physical Sciences 349
- Earth and Environment 26
- Life Sciences 26
Start searching
38,465 projects, 2002–2026, from 6 competitions.