School districts track attendance, assessment results, and graduation rates closely. Very few track the air students breathe while those results are being produced.
A growing body of research suggests they should. In an August 2026 article, AirGradient's Ethan Brooke reviews the latest evidence connecting air pollution to academic outcomes. His conclusion is that the research has moved past "this probably matters." It now produces figures a facilities director can bring to a budget meeting.
This post summarizes the key findings from that review and the studies it cites. It then looks at what they mean for the people who run school buildings.
The evidence, in brief
Fine particles and decision quality
Brooke opens with a 2023 study of professional chess tournaments. Researchers matched indoor air quality readings in tournament halls against more than 30,000 moves, each graded by a chess engine. For every 10 µg/m³ increase in indoor PM2.5, players were 26.3% more likely to make a meaningful error. The effect was stronger under time pressure.
The concentrations in those halls ranged from roughly 14 to 70 µg/m³. As the source notes, those are levels found in many ordinary classrooms.
Smoke days and standardized test scores
A Stanford-led analysis covered nearly 11,700 US school districts between 2009 and 2016. It matched test results against satellite estimates of wildfire smoke. Math and English scores fell in smokier school years. The effect was concentrated in grades 3 to 5, and it was nearly twice as large when smoke arrived on school days rather than weekends.
Everyday traffic pollution
Most schools will rarely sit under wildfire smoke. A study of students who changed schools compared those who moved to a school downwind of a major highway with those who moved to an upwind school at a similar distance. Students at downwind schools had lower test scores, more absences, and more behavioral incidents. The main difference between the two groups was which way the wind carried traffic pollution.
The takeaway is that routine, background pollution affects students, not just extreme events.
Exam-day air and lifetime earnings
An Israeli study of high-stakes Bagrut exit exams compared the same students' scores on cleaner and more polluted exam days. Scores dropped on high-PM2.5 days. Because those exams shape university admission, the effect lasted. About a decade later, students who sat exams in worse air were earning roughly USD 30 per month less than peers who tested on cleaner days.
The closed-window dilemma
The obvious response to outdoor PM2.5 is to close the windows. Brooke points out why that creates a second problem.
Classrooms are among the most densely occupied rooms in daily life, with 25 to 30 people sharing one space for hours. Ventilation surveys across multiple countries often find classrooms below recommended rates, at around 3 to 5 liters per second per person. Peak CO₂ during lessons regularly passes 2,000 ppm, and maximum readings range from 3,000 to 6,000 ppm.
Exhaled CO₂ also serves as an indicator of poor ventilation. The source cites a 2016 Harvard chamber study in which cognitive test scores at about 1,400 ppm CO₂ were roughly half those recorded on the cleanest-air days. Brooke is careful to note that the study had only 24 participants and that later CO₂-only research has found smaller, less consistent effects. However, a six-country field study in real offices found the same direction of effect. As ventilation fell, response times slowed and accuracy dropped, and PM2.5 had a separate impact on top of that.
In a naturally ventilated classroom, the dilemma is real:
- Open the windows on a polluted day, and outdoor PM2.5 comes in.
- Close them, and CO₂ from 30 occupants builds up quickly.
As the source puts it, teachers end up choosing between pollutants, usually with no way to tell which one is the bigger problem at that moment.
What the interventions show
The strongest evidence that cleaner air improves learning comes from an unplanned experiment. After the 2015 Aliso Canyon gas leak in Los Angeles, air filters were installed in every classroom within five miles of the site. Comparing schools just inside and just outside that boundary, a researcher found filtration raised math scores by 0.20 standard deviations, which is roughly three months of additional learning, and English scores by 0.18. No gas was detected inside the schools. The filters were removing everyday urban pollution.
Filtration alone only solves half of the problem, though. A recirculating purifier can reduce PM2.5, but it does nothing about CO₂. Brooke argues that positive-pressure systems, which bring in filtered outdoor air and keep the room slightly pressurized, can address both at once. They dilute CO₂, filter particles on the way in, and push leakage outward so unfiltered air does not seep in. For open-air or semi-open classrooms that cannot be sealed, he suggests setting up at least one sealed, filtered clean room as a refuge on the worst days.
We covered why positive-pressure performance must be verified rather than assumed in a previous post.
The AirSafe Intel angle: you can't manage the dilemma without data
The research makes one point clear: the right action depends on conditions in a specific room at a specific time. Should the windows be open or closed? Is the purifier enough, or is CO₂ the bigger risk right now? Is the clean-air investment actually improving the air students breathe?
Without measurement, facility teams and teachers are guessing. Continuous monitoring turns those questions into decisions.
1. Show which pollutant is the problem, room by room. The same building can have one classroom with high PM2.5 from a nearby road and another with high CO₂ from poor airflow. Monitoring CO₂, PM2.5, VOCs, and temperature together shows which problem each room has, instead of applying one fix everywhere.
2. Give teachers a clear signal, not a judgment call. The closed-window dilemma exists because the person in the room can't see either pollutant. A single 0–100 Air Risk Score (green ≤ 40, amber 41–70, red > 70) and automated alerts give staff a simple cue to act on. Examples include opening windows, starting ventilation, or moving students to a filtered space.
3. Protect the hours that matter most. The Stanford and Bagrut findings suggest that timing matters. Pollution during instruction and exams does the most damage. Trend data shows whether rooms used for testing, early grades, or long instructional blocks regularly exceed ventilation targets such as those in ASHRAE 62.1 during occupied hours.
4. Prove that interventions are working. The Aliso Canyon result shows that filtration can pay off academically. Districts investing in purifiers, positive-pressure units, or HVAC upgrades should be able to show before-and-after data for the rooms they changed. That same data supports WELL and LEED documentation and capital planning conversations with boards.
5. Find the rooms no one is looking at. Most schools don't know which classrooms are their worst performers. Portfolio-wide visibility shows district leaders where students are exposed and where the next dollar of IAQ spending will do the most good.
From awareness to accountability
Brooke ends his review by arguing that the conversation needs to move beyond awareness, and we agree. The evidence now links classroom air to measurable outcomes: fewer errors, higher test scores, and stronger long-term prospects for students.
For school operators, the practical next step is knowing what is happening in every classroom, all day long. That visibility separates a clean-air policy from clean air students can actually rely on.
AirSafe Intel is working with early-access school and district partners to deploy continuous CO₂, PM2.5, VOC, and temperature monitoring, with real-time alerts and a unified Air Risk Score across every monitored space.
Primary source: Ethan Brooke, “Air Pollution Is Costing Students More Than Marks,” AirGradient, August 21, 2026. The research findings summarized above, including the chess, wildfire smoke, traffic, Bagrut, ventilation survey, Harvard, six-country office, and Aliso Canyon studies, are attributed to that article and the original publications it links. The recommendations in "The AirSafe Intel angle" section are AirSafe Intel's own analysis.