100+ Best STEM Science Fair Project Ideas for High School Students
Most students approach a science fair the wrong way. They pick a topic they already know something about, find a project someone else has done before, tweak one variable, and call it original research. The result is a project that looks fine on a poster board and goes nowhere competitively, because the judges have seen it, or something almost identical to it, dozens of times.
Whether you're aiming for a regional fair, Regeneron ISEF, or just want to produce something you're genuinely proud of, we’ve compiled a list of fair ideas. They're specific enough to execute, open enough to produce original findings, and grounded in domains where you can do work that matters.
Before diving into the list: if you're unsure how to approach scoping, executing, or presenting a STEM research project, a mentored research program like Veritas AI can make a huge difference! Working with a mentor who has done this kind of research themselves takes a project from "I built something" to "I can defend every decision I made."
How can a STEM fair project help me with college admissions?
A strong STEM fair project also does more than win a ribbon or a cash prize. The research skills you develop, designing experiments, handling real data, writing up methodology, presenting to skeptical judges, are exactly what college professors and lab supervisors look for in incoming students. Admissions readers at competitive universities can quickly understand the difference between a student who’s listed "science fair participant" and one who can describe their methodology and findings in a 5-minute conversation.
P.S. We've also put together a list of top high school science fair project ideas if you want more inspiration, a guide to the best science competitions for high school students to find the right stage for your work, and a breakdown of STEM research paper competitions for when you're ready to take your project all the way to publication.
What do STEM fair judges look for in an idea?
Before the ideas, here’s a quick primer on what separates a finalist from the rest of the field. Most judges across major STEM fairs score on five dimensions: scientific thought, creative ability, thoroughness, skill, and clarity of presentation. The first two matter most.
Scientific thought means your project has a real hypothesis, a method capable of testing it, and conclusions that follow from your data, not from what you hoped to find. Creative ability means the question itself is non-obvious. You didn't Google "good science fair projects" and pick the top result. You identified a gap or an unexplored angle in a domain you care about.
The best STEM fair projects tend to be narrow. Not "how does social media affect mental health" but "does Instagram use predict differences in self-reported anxiety between students who primarily consume vs. primarily post content?" The narrower the question, the stronger the methodology, and the more defensible the conclusions!
With that, here are 100+ great STEM fair ideas organized by the domain!
AI and Machine Learning STEM Fair Projects
AI/ML projects have become one of the strongest categories at major science fairs because the tools are accessible, the datasets are public, and the applications are genuinely impactful. The key is going beyond the tutorial and asking a question that hasn't been cleanly answered.
Predicting student dropout risk using a gradient boosting model trained on attendance and grade data from a public school district dataset
Comparing the diagnostic accuracy of a CNN trained on dermoscopy images against published dermatologist benchmark scores
Detecting early-stage Parkinson's disease from voice recordings using MFCC features and an SVM classifier
Auditing a recidivism prediction algorithm (like COMPAS) for demographic bias using a fairness metrics framework
Building a sign language gesture recognizer using MediaPipe and measuring accuracy across different skin tones
Comparing the performance of transformer-based vs. traditional NLP models on low-resource language classification
Measuring the carbon cost (energy per inference) of different model architectures on equivalent tasks
Predicting wildfire spread risk using weather, terrain, and historical fire perimeter data
Building a misinformation detector and evaluating its precision and recall on a labeled dataset
Using transfer learning to classify rare species from camera trap images and measuring how training set size affects accuracy
Comparing ML approaches to predicting hospital readmission risk and evaluating clinical utility vs. existing benchmarks
Studying how adversarial examples fool image classifiers and measuring the robustness of different defense techniques
Environmental Science STEM Fair Projects
Environmental projects tend to perform well when they're locally grounded and data-rich. Judges respond to projects where the student clearly cares about the place they're studying, and where the findings have a real policy or management implication.
Measuring microplastic concentration in local waterways at different distances from urban runoff sources
Studying the effect of road salt application on soil salinity and plant germination rates near treated roads
Comparing pollinator visit frequency to native vs. non-native plant species in a suburban garden setting
Measuring the urban heat island effect in your city using low-cost temperature sensors placed at varying impervious surface densities
Studying the effect of light pollution intensity on insect behavior using a controlled light trap experiment
Analyzing the relationship between tree canopy coverage and ambient temperature in urban neighborhoods using public GIS data
Measuring the effectiveness of different biochar types as soil carbon sequestration agents
Studying whether rain garden design affects the rate of heavy metal removal from stormwater runoff
Comparing dissolved oxygen levels upstream and downstream of agricultural drainage outlets in a local watershed
Measuring the decomposition rate of different "compostable" plastics under realistic home composting conditions
Analyzing how invasive plant species affect soil microbial diversity using a public metagenomics dataset
Studying the relationship between impervious surface coverage and stream flashiness in urban watersheds
Biomedical and Health Science STEM Fair Projects
These projects require careful ethics consideration (especially anything involving human subjects) but are consistently strong at major competitions because the stakes are clear and the methodology is rigorous.
Comparing the antibacterial effectiveness of commercial hand sanitizers vs. soap against a non-pathogenic bacteria proxy
Studying the effect of blue light wavelength on melatonin suppression using a validated sleep quality survey instrument
Analyzing whether grip strength is a reliable predictor of overall cardiovascular fitness in high school athletes
Comparing the accuracy of consumer-grade wearables in measuring heart rate variability during different activity types
Studying the relationship between gut microbiome diversity and self-reported anxiety using a publicly available dataset (HMP or similar)
Measuring the effect of different music tempos on reaction time and working memory performance
Analyzing the effect of acute dehydration on fine motor control using a standardized steadiness test
Studying whether a brief mindfulness intervention measurably affects cortisol proxy markers (salivary alpha-amylase)
Comparing vision board vs. implementation intention planning methods on short-term goal completion rates (student-designed RCT)
Analyzing public ER wait time data to identify the demographic and geographic predictors of delayed care access
Measuring how different types of dietary fiber affect in vitro fermentation rates using a simulated gut model
Studying the relationship between adolescent sleep duration and academic performance using a longitudinal survey design
Physics and Engineering STEM Fair Projects
Measuring how blade geometry affects the power output of a small-scale wind turbine at varying wind speeds
Studying the effect of surface texture on the drag coefficient of 3D-printed objects using a water channel
Comparing the energy harvesting efficiency of piezoelectric materials under different mechanical stress configurations
Measuring how the angle and spacing of solar panels affect total energy output under varying cloud cover conditions
Studying the acoustic properties of different recycled materials as sound insulation
Comparing the structural load tolerance of different 3D-printed infill patterns at equivalent material weight
Measuring the effect of water temperature and salinity on the efficiency of a simple desalination prototype
Studying how varying the geometry of a passive solar greenhouse affects internal temperature distribution
Comparing the efficiency of different thermoelectric generator configurations for waste heat recovery
Measuring vibration damping properties of different bio-based composite materials compared to conventional foam
Studying the relationship between bridge truss geometry and load distribution using scale models and force sensors
Designing and testing a low-cost, passive water collection system optimized for fog-prone climates
Chemistry STEM Fair Projects
Studying the rate of microplastic release from different synthetic fabric types during simulated laundry cycles
Comparing the effectiveness of natural vs. synthetic mordants on dye uptake and wash-fastness in textile dyeing
Measuring the effect of pH on the rate of enzymatic browning in different apple varieties
Studying the photocatalytic degradation of methylene blue dye using zinc oxide nanoparticles under UV and visible light
Comparing the corrosion resistance of different metal alloys under varying concentrations of acidic and saline solutions
Measuring the effect of temperature cycling on the structural integrity of biodegradable packaging materials
Studying whether spent coffee grounds can effectively remove heavy metal ions from aqueous solution
Comparing the energy density and discharge characteristics of different homemade electrochemical cell configurations
Measuring how extraction method and solvent polarity affect the antioxidant yield from common plant sources
Studying the effect of surfactant concentration on the stability and droplet size of an emulsion
Computer Science and Data Science STEM Fair Projects
Studying whether graph centrality metrics can predict information spread in a simulated social network
Comparing the performance of different sorting algorithms on real-world vs. random datasets of varying structure
Measuring how training data diversity affects the accuracy and fairness of a face recognition model
Studying the effect of hyperparameter tuning on overfitting in small neural networks using learning curve analysis
Analyzing traffic accident patterns in public city data to identify infrastructure features correlated with higher incident rates
Comparing the compression efficiency and reconstruction quality of different image compression algorithms
Studying whether sentiment in a city's social media correlates with publicly available wellbeing and economic indicators
Measuring how different randomization methods in A/B testing affect the minimum detectable effect size
Building a network intrusion detection classifier and testing it against labeled public cybersecurity datasets
Studying the mathematical properties of different graph traversal algorithms on networks with varying density and clustering
Biology and Life Sciences STEM Fair Projects
Comparing the germination rate and early growth of seeds exposed to different magnetic field strengths
Studying the effect of caffeine concentration on Daphnia (water flea) heart rate as a model organism
Measuring how different light wavelengths affect the rate of photosynthesis in aquatic plants using oxygen evolution
Comparing the allelopathic effects of extract from an invasive plant species on the germination of local native plants
Studying the effect of microbiome-derived short-chain fatty acids on the behavior of C. elegans (using existing datasets)
Measuring the effect of substrate moisture variability on mycelium growth rate and network density
Comparing the feeding selectivity of local pollinator species across native vs. cultivated flower varieties
Studying the relationship between leaf morphology and photosynthetic efficiency across sun-exposed and shade-adapted plants
Measuring the antimicrobial activity of different concentrations of essential oils against a non-pathogenic bacteria model
Analyzing publicly available eDNA metabarcoding data to study seasonal biodiversity shifts in a local ecosystem
Mathematics and Statistics STEM Fair Projects
Testing whether Benford's Law holds in multiple real-world datasets and analyzing why it breaks down in some cases
Studying the predictive accuracy of a game theory model on outcomes in a specific competitive sport
Modeling the spread of a historical disease outbreak using an SEIR model and comparing the fit to actual case data
Analyzing the distribution of prime gaps computationally and testing whether patterns hold across ranges of increasing magnitude
Studying the accuracy of different polling aggregation methods in predicting outcomes vs. single-poll models
Comparing the convergence rate of different numerical integration methods on functions with varying smoothness
Applying network centrality analysis to co-authorship data in a scientific field to identify influential researchers
Studying the relationship between a city's street network topology and average commute times using public data
Modeling opinion polarization using agent-based simulation and testing the effect of network structure on final consensus
Analyzing the mathematical properties of different ranking systems (Elo, Glicko, TrueSkill) on a synthetic tournament dataset
Psychology and Cognitive Science STEM Fair Projects
These require careful study design and IRB-equivalent ethics review at most major fairs, but they're consistently distinctive.
Studying the effect of background noise type (white noise, music, ambient) on reading comprehension and recall
Comparing the accuracy of eyewitness recall under high vs. low cognitive load conditions using a standardized memory protocol
Measuring whether implicit bias scores (IAT) predict explicit behavior in a controlled social preference task
Studying the effect of growth mindset framing on persistence after failure in a puzzle-solving task
Comparing the decision quality of intuitive vs. analytical thinking styles on problems with varying structure
Measuring the effect of physical workspace organization on creative output in a divergent thinking task
Studying whether priming with scarcity cues affects risk tolerance in a simulated economic decision task
Comparing the memorability of information presented as narrative vs. list format across different age groups
How to Turn Any of These Into a Winning Project
The idea is only about 10% of a strong science fair project. The other 90% is methodology, execution, and writeup. A few things that consistently separate finalists from the rest:
Your hypothesis has to be falsifiable. If there's no result that would prove you wrong, you don't have a hypothesis, you have a demonstration. Every project on this list can be framed as a genuine test of something uncertain.
Controls matter more than most students realize. Every variable you don't control for is a potential alternative explanation for your results. Identifying those alternative explanations and addressing them, either by controlling them or by acknowledging them honestly in your limitations section, is what scientific rigor actually looks like.
Negative results are valid. If your model performs worse than the baseline, that's a finding. If your intervention had no measurable effect, that's data. Judges who understand research respect honest null results far more than they respect massaged conclusions.
Your writeup should read like a paper, not a poster. Even if you're presenting on a board, the underlying document should have a proper abstract, introduction, methodology, results, discussion, and bibliography. Projects that look like they could be submitted to a journal stand out immediately.
Do I need mentorship for a STEM fair?
The gap between a competent science fair project and a genuinely competitive one almost always comes down to mentorship. A mentor who has done research in your domain can help you identify the non-obvious question, flag the methodological issues before you run your experiment, and push back on conclusions that don't follow from your data. That kind of feedback is hard to replicate from a textbook or a YouTube video.
Veritas AI is built around exactly this model. High school students work directly with mentors from top universities and AI companies to develop and execute original AI research projects, with a structured curriculum covering machine learning, Python, data science, model evaluation, and AI ethics. The program culminates in a completed capstone project and presentation, with a track record of student work reaching publication and conference submission. If your STEM fair project sits at the intersection of data science, machine learning, or AI, this is the mentorship infrastructure that gives your work real depth.
Frequently Asked Questions
What makes a good STEM science fair project for high school? A specific, answerable research question, a methodology with real controls, original data or analysis, and honest conclusions. The most important thing is that something was genuinely unknown before you started, and your project produced evidence relevant to answering it.
What are the best STEM fair categories for high school students? AI and machine learning, environmental science, and biomedical science are consistently strong categories at major fairs because the questions are current, the datasets are often publicly available, and the real-world stakes are clear. That said, the best category is the one where you have the deepest genuine interest, because that shows in the quality of the work.
How do you find a mentor for a science fair project? University lab websites, cold emailing professors with a specific research question, science fair mentorship programs, and structured programs like Veritas AI are all legitimate paths. The more specific your question, the more likely a researcher in the relevant field will be willing to give you time.
Can a high school student publish science fair research? Yes. Journals like the Journal of Emerging Investigators are specifically designed for high school research. More competitive venues like Regeneron ISEF and JSHS also have tracks that result in recognition and, in some cases, publication opportunities. Work done under structured mentorship programs tends to meet publication standards more consistently.
How early should you start a STEM fair project? Most strong projects require at least three to four months of active work, and competitive projects at the Regeneron ISEF level often represent six or more months of effort. Starting in the summer before the fair year is ideal. The projects that run out of time are almost always the ones that started too late to iterate on their methodology after the first results came in.