20 Science Fair Project Ideas for High School Students (2026)
20 Science Fair Project Ideas for High School Students (2026)
There are two kinds of science fair projects, and judges can tell them apart in under a minute.
The first demonstrates something already known. A working model of a wind turbine. A solar oven that boils water. A volcano. These are fine in Class 7 and slightly embarrassing in Class 11, because nothing was actually tested; the answer was in the textbook before the student started.
The second asks a question nobody at the fair knows the answer to, then measures its way to one. That's the category that wins, and the surprising part is that it rarely costs more. What it needs is a real variable, a control, enough repeats to survive a statistics question, and a method the student can defend when a judge pushes back.
The twenty science fair ideas below are all of the second kind. Each is genuinely achievable by a high school student in India without access to a university lab, and each has a defensible answer at the end rather than a demonstration. A few involve microbial culture or dilute acid, and those need a teacher to supervise and a plan for disposal that's noted where applicable.
Want even more inspiration? Explore our Project Ideas Generator and find the project idea perfect for you
Biology and microbiology
1. Screen your local soil for bacteria that kill other bacteria
Most clinically important antibiotics were discovered in soil microbes, and the search is nowhere near finished. Dilute soil from contrasting sites, spread it on nutrient agar, and hunt for colonies surrounded by clear halos where neighbouring growth has been inhibited. Isolate those producers, challenge each against a harmless indicator strain, and measure inhibition zones in millimetres.
Field: Microbiology · Difficulty: Hard · Timeline: 6–8 weeks
Equipment: nutrient agar, petri dishes, inoculation loop, pressure cooker for sterilising, soil from several sites, ruler, supervising teacher.
2. Does storing drinking water at home make it less safe?
Most Indian households store water between supply hours, and almost nobody asks what storage time does to it. Sample the same source at collection and again after 6, 24, and 48 hours across container types, steel, plastic, clay, covered and uncovered, testing each with H2S vials or a coliform presence-absence kit. Report contamination rate against material and duration.
Field: Public health microbiology · Difficulty: Intermediate · Timeline: 3–4 weeks
Equipment: H2S test vials or coliform kit, sterile sample bottles, containers of different materials, thermometer.
3. Find the point where sugar stops helping yeast and starts killing it
Fermentation speeds up with sugar until osmotic stress starts dehydrating the cells, which means there's a real optimum and a real collapse. Run identical yeast suspensions across eight sugar concentrations, collect the carbon dioxide in an inverted cylinder over water, and log gas volume every two minutes. Fit the curve and identify the inhibition threshold.
Field: Biochemistry · Difficulty: Intermediate · Timeline: 2–3 weeks
Equipment: baker's yeast, sugar, measuring cylinder, water trough, delivery tubing, conical flasks, thermometer, stopwatch.
4. Do invasive trees poison the ground against native seeds?
Eucalyptus, Prosopis juliflora and lantana are all suspected of allelopathy, leaking compounds that suppress competitors' germination. Prepare leaf extracts at five concentrations, apply them to native seeds on filter paper, and measure germination percentage and radicle length against a water control. A dose-response curve is far harder to dismiss than a single comparison.
Field: Plant ecology · Difficulty: Intermediate · Timeline: 3–4 weeks
Equipment: leaves of the invasive species, native seeds, filter paper, petri dishes, blender, weighing scale, ruler.
Chemistry and materials
5. Put your kitchen through real food adulteration tests
FSSAI publishes simple field tests for common adulterants, but nobody quantifies how often they actually turn up. Buy the same product across vendor types- loose, local brand, national brand- and run standard tests for metanil yellow in turmeric, starch and detergent in milk, and chalk in flour. Photograph every result under fixed lighting and score colour intensity in ImageJ, turning a yes/no test into a number.
Field: Analytical chemistry · Difficulty: Intermediate · Timeline: 3–4 weeks
Equipment: samples from multiple vendors, dilute hydrochloric acid, iodine solution, test tubes, phone camera, a fixed lightbox, ImageJ.
6. Turn a phone camera into a colorimeter, and calibrate it honestly
A lab colorimeter is expensive; a phone in a cardboard lightbox does the same job if the calibration is done properly. Photograph a series of known-concentration standards of a colour-forming reaction, extract mean RGB values, and regress concentration against whichever channel responds most strongly. Then measure unknown samples for hardness, nitrate or iron, and validate against commercial strips.
Field: Instrumentation · Difficulty: Hard · Timeline: 4–6 weeks
Equipment: cardboard lightbox, phone with exposure lock, reagent kit for one analyte, volumetric glassware, standards, ImageJ or Python.
7. Which plant extract protects steel best, and by exactly how much?
Green corrosion inhibitors are an active research area, and inhibition efficiency has a clean formula a student can calculate. Weigh identical mild steel coupons, immerse them in dilute acid with and without extracts of neem, tea, onion peel, or tulsi, then reweigh after a fixed immersion. Efficiency is the percentage reduction in mass loss against the uninhibited control.
Field: Materials chemistry · Difficulty: Intermediate · Timeline: 3–4 weeks
Equipment: mild steel coupons, dilute hydrochloric acid, plant material, scale reading to 0.01 g, beakers, gloves and goggles.
8. Make biochar from crop waste and measure what it strips from water
Rice husk, coconut shell and groundnut shell all char into serviceable adsorbents, and adsorption obeys laws you can actually fit. Char your waste in a covered tin, grind and sieve it, then shake fixed doses with dye solution across a range of contact times. Measure the colour remaining and fit your results to a Langmuir or Freundlich isotherm.
Field: Environmental chemistry · Difficulty: Hard · Timeline: 5–6 weeks
Equipment: agricultural waste, tin with lid and heat source, mortar, sieve, methylene blue, conical flasks, colorimeter setup.
Physics and engineering
9. Measure Planck's constant with a handful of LEDs
An LED begins emitting at the voltage where electron energy matches photon energy, which means five coloured LEDs and a multimeter can deliver a fundamental constant of nature. Record threshold voltage for LEDs of known wavelength, plot voltage against frequency, and take the gradient. You should land within roughly ten percent, and explaining that gap is the real work.
Field: Quantum physics · Difficulty: Intermediate · Timeline: 2–3 weeks
Equipment: five LEDs of known wavelength, breadboard, potentiometer, multimeter, battery, dark box.
10. Build a PM2.5 monitor and check it against the government's
Low-cost particulate sensors are everywhere now, and almost nobody validates them. Wire a laser dust sensor to a microcontroller, log PM2.5 every minute for a fortnight at a fixed spot, then pull matching hourly readings from the nearest CPCB station. Regress the two series, quantify the drift, and test whether humidity explains the error.
Field: Environmental engineering · Difficulty: Hard · Timeline: 5–7 weeks
Equipment: ESP32 or Arduino, PMS5003 or SDS011 sensor, humidity sensor, SD card module, power supply, CPCB public data.
11. Find out how much bounce a ball loses when it gets cold
Coefficient of restitution is one clean number, and phone slow-motion at 240 frames per second measures it precisely enough. Condition balls of several materials in a fridge, at room temperature and in direct sun, drop each from a fixed height, and read rebound heights frame by frame in Tracker. Plot restitution against temperature for every material.
Field: Mechanics · Difficulty: Easy · Timeline: 2 weeks
Equipment: balls of different materials, phone with slow-motion, tripod, metre scale, thermometer, fridge, Tracker software.
12. Work out when an evaporative cooler beats a fan, and when it fails
Evaporative cooling depends entirely on how dry the air is, which is why desert coolers are excellent in May and useless in August. Build a small cooler, run it while logging dry-bulb temperature, wet-bulb temperature, and relative humidity, and calculate cooling efficiency across conditions. Then find the humidity at which the advantage disappears altogether.
Field: Thermal engineering · Difficulty: Intermediate · Timeline: 4 weeks
Equipment: khus pads or clay pot, small fan, dry and wet bulb thermometers, hygrometer, wick or pump.
Environment and earth science
13. Map your city's heat islands from free satellite data
Landsat carries a thermal band, so land surface temperature and vegetation index are both free to anyone willing to learn the tools. Pull cloud-free summer scenes for your city, compute NDVI and surface temperature in QGIS or Google Earth Engine, and sample both at fifty points across built-up areas, parks, and water bodies. Regress temperature against greenness to quantify cooling per unit of canopy.
Field: Remote sensing · Difficulty: Hard · Timeline: 5–6 weeks
Equipment: a computer, QGIS or Google Earth Engine, Landsat imagery, city ward boundaries.
14. Count the microplastics in what you eat and drink
Microplastic contamination of Indian salt, sugar and bottled water is documented, and the counting method is within a school's reach. Filter dissolved samples through fine paper, dry them, scan under a stereo microscope, and confirm suspect particles with the hot-needle test. Report particles per kilogram by brand category, comparing packaged against loose.
Field: Environmental science · Difficulty: Intermediate · Timeline: 4–5 weeks
Equipment: stereo or digital microscope, fine filter paper, filtration setup, glass beakers, needle, distilled water.
15. Test whether the first rain off your roof is worth discarding
Harvesting guidelines say to throw away the first flush, but the right volume depends on roof material and how long the dry spell was. Collect successive fixed volumes as rain runs off, testing each for turbidity, pH, TDS and coliforms, and record the preceding dry days. Determine the discard volume beyond which quality stops improving.
Field: Water resources · Difficulty: Intermediate · Timeline: one monsoon month
Equipment: collection bottles, TDS meter, pH meter or strips, turbidity tubes, coliform vials, rain gauge.
16. Build a noise map of your neighbourhood and test it against WHO limits
Urban noise is a recognised health risk and is rarely mapped at street level. Calibrate a phone sound-level app against a borrowed reference meter, then log readings at fifty geotagged points at three fixed times of day across a week. Plot a heat map and calculate what share of residential points exceed WHO guideline levels.
Field: Environmental health · Difficulty: Easy · Timeline: 2–3 weeks
Equipment: phone with sound-level app, borrowed reference meter, GPS, QGIS or Google My Maps.
Data science, computing and psychology
17. Train a leaf disease classifier, then find out where it breaks
Anyone can reach 95% accuracy on the public PlantVillage dataset. The actual finding is how far that collapses on photographs you take yourself, in real light, against real backgrounds. Train the model, test it on your own images, and build a confusion matrix of failure modes. Report the accuracy gap and diagnose what causes it.
Field: Machine learning · Difficulty: Hard · Timeline: 5–6 weeks
Equipment: a computer, Python with TensorFlow or Teachable Machine, PlantVillage dataset, phone camera, access to real plants.
18. Does air quality track hospital visits in your district?
CPCB publishes air quality daily, and many state health portals publish outpatient counts, which makes a real epidemiological study possible without collecting a single sample. Assemble two years of both series, test correlation at lags of zero to seven days, and control for temperature and season. Discuss confounders carefully instead of claiming causation.
Field: Data science · Difficulty: Intermediate · Timeline: 4 weeks
Equipment: CPCB data, district health statistics, spreadsheet or Python with pandas, basic statistics.
19. Change the wording, change the answer: run a framing experiment
Framing effects are among the most reproducible findings in behavioural science, and replicating one rigorously is a legitimate project. Write two versions of a decision scenario that are mathematically identical but phrased as a gain or a loss, randomly assign 120 respondents to one version, and compare choice proportions with a chi-square test. Write your hypothesis down before you collect anything.
Field: Behavioural psychology · Difficulty: Intermediate · Timeline: 3–4 weeks
Equipment: survey form, 120 respondents, randomisation method, spreadsheet, chi-square calculator, consent statement.
20. Audit how often an AI chatbot gets your syllabus wrong
Students everywhere now check their homework against chatbots, and there is almost no classroom-level evidence on how reliable that is. Build a marking rubric, run 100 questions from your board syllabus through two or more models, and have a second person score independently so you can report inter-rater agreement. Classify every error by type arithmetic, conceptual, fabricated source and by chapter.
Field: Educational technology · Difficulty: Easy · Timeline: 3 weeks
Equipment: access to two AI chatbots, 100 syllabus questions with verified answers, scoring rubric, a second scorer.
Where to take a high school project in India in 2026
The strongest route for research-grade work is the IRIS National Fair, India's affiliate of the Regeneron International Science and Engineering Fair; winners go on to represent India at ISEF. Separately, if you're in classes 6 to 12, your school can nominate your idea for INSPIRE Awards–MANAK from the Department of Science & Technology, which funds shortlisted ideas and runs district, state and national rounds. Nominations go through your principal, not students directly. NCERT also runs a national science exhibition feeding up through state rounds. And students with a written paper rather than a poster can present at ICONSTARS 2026, an international student research conference on 8–9 August 2026.
When a science fair project becomes publishable research
Several of these science fair ideas are one step away from a paper. The satellite heat island study, the low-cost sensor validation, the adsorption isotherms, the framing replication- all of them produce original data on questions that are genuinely open. What usually stops a student is not ability but structure: no literature review, no methodology they can defend to a reviewer, and nobody to tell them which findings matter.
That's the gap the Global R-Hub High School Scholar Program exists to close. Over three to six months, a PhD mentor works with a Class 9–12 student to turn a question into a publishable 10–15 page paper, literature review, methodology, analysis, and submission to a peer-reviewed journal, with global faculty reviewing the work. Students finish with a published paper and a letter of recommendation before university applications open.
If your child already has a project worth defending, that's the point at which mentorship compounds. Start with the free 2026 Research & Admissions Guide, or book a free 45-minute strategy session to map what their paper could look like.
