Yes, radon levels change over time, and they can shift dramatically—by a factor of two to ten or more—across hours, days, seasons, and years. Radon is a radioactive gas produced by uranium decay in soil, and its concentration in any building depends on a stack of variables that never hold still: soil moisture, barometric pressure, temperature gradients, ventilation habits, and the building's own pressure balance. A single short-term test is a snapshot, not a verdict. Understanding why the number moves is the difference between a meaningful measurement and a false sense of safety.
The transport mechanism: why radon rarely sits still
Radon-222 forms when radium-226 decays in soil and rock. It is a noble gas, so it does not bind chemically to soil particles; instead it migrates through pore spaces and fractures, driven by pressure differences between the soil and the building above it. This is the part most explainers skip: radon entry is fundamentally a pressure-driven process, not a diffusion process. Diffusion moves radon centimeters through dense soil over days; pressure-driven advection moves it meters through cracks in minutes.
A house behaves like a low-pressure zone relative to the soil beneath it. Warm indoor air rises through the stack effect, leaks out the upper envelope, and the lower floors draw replacement air inward—including soil gas through foundation cracks, sump pits, slab penetrations, and drain tiles. Anything that changes that pressure balance changes the radon entry rate.
Stack effect strength scales with the indoor-outdoor temperature difference. A 40°F indoor-outdoor gap in January drives far more suction at the foundation than a 5°F gap in May. This single mechanism explains most of the seasonal pattern discussed below.
The time course: hours to decades
Radon concentration is a moving target at every timescale. The table below maps the dominant drivers at each scale.
| Timescale | Typical variation | Dominant driver |
|---|---|---|
| Hourly | 2x to 5x swings | Pressure changes, wind, appliance exhaust |
| Daily | 2x to 10x swings | Temperature gradient, weather fronts |
| Weekly | 2x to 5x swings | Soil moisture, rain saturation |
| Seasonal | 2x to 5x (winter high in cold climates) | Stack effect, frozen ground, closed windows |
| Year-to-year | 20% to 50% drift | Soil settling, foundation changes, HVAC changes |
| Decadal | Structural aging, soil redistribution | Foundation degradation, major renovations |
Hourly swings deserve emphasis because they defeat the intuition that "the number is the number." A closed house with a running furnace and a bathroom fan can pull radon from 2 pCi/L to 8 pCi/L in a few hours, then back down when the fan stops. The EPA action level of 4 pCi/L is a long-term average benchmark, not a line that a single reading crosses or clears.
Seasonally, cold-climate homes typically peak in winter: frozen ground caps soil gas escape at the surface, snow cover seals the soil, windows stay closed, and the stack effect runs at maximum. Warm-climate homes can invert this pattern because air conditioning and closed windows dominate. The counterintuitive detail: a summer reading can understate annual exposure by a factor of three in a northern home, which is why short-term tests carry explicit seasonal caveats.
What moves the number: seven variables that matter
Soil moisture is the most underappreciated variable. Rain saturates pore spaces, temporarily blocking gas flow at the surface and pushing more radon toward the foundation; several days later, as soil dries and cracks open, entry can spike again. Barometric pressure works inversely: falling pressure draws soil gas into the house, rising pressure suppresses it. A passing storm front can move indoor radon by 50% within hours.
Mechanical systems are the second major lever. Furnaces, dryers, range hoods, and bath fans all depressurize the house and increase soil gas entry. A 200-cfm bath fan in a tight house can measurably raise radon within an hour. Conversely, a balanced ventilation system or an open window dilutes indoor radon—but only while it runs, which is why "I opened the windows and the test was fine" is a classic measurement error.
The measurement problem: why your two tests disagree
Radon measurement devices do not measure radon directly. They count decay events—alpha particles from radon progeny—and convert counts to concentration using calibration factors. Short-term charcoal canisters integrate over 2 to 7 days and are sensitive to placement, closed-house conditions, and seasonal timing. Continuous monitors log hourly data and reveal the swings, but their accuracy depends on calibration and their placement can be biased by proximity to the foundation.
The professional caveat that separates serious analysis from thin explainers: two consecutive short-term tests in the same room can legitimately differ by 100% or more, and neither is "wrong." The EPA recommends confirmatory testing rather than treating a single result as definitive, and it distinguishes between short-term screening (2 to 90 days) and long-term testing (more than 90 days, up to a year) for the most reliable annual average. If your first test reads 6 pCi/L and a retest reads 2.5 pCi/L, the honest interpretation is "annual average is uncertain; test longer or mitigate based on the higher result."
Thresholds, reconciled
Multiple numbers circulate, and they serve different purposes. The EPA action level is 4 pCi/L—the level at which mitigation is recommended. The EPA also notes that 2 pCi/L is a "consider action" zone because no level is entirely risk-free. The WHO reference level is 100 Bq/m³, which converts to roughly 2.7 pCi/L. These are not contradictory; they reflect different risk tolerances and averaging periods. The 4 pCi/L action level is a long-term average, not a single-reading trigger, so a one-day spike to 8 pCi/L is not automatically a mitigation mandate—but it is a signal to test longer.
Red flags that demand immediate attention, not more testing:
A long-term test averaging above 4 pCi/L in a lived-in space. Radon above 4 pCi/L in a basement bedroom or any occupied lower level. A short-term test above 8 pCi/L, which warrants confirmatory long-term testing promptly. Any detectable radon in a home with an infant or a smoker, where combined risk compounds. A mitigation system that has never been retested since installation.
Practical actions: how to measure a moving target
Test long enough to average out the noise. The most defensible protocol for a homeowner is a long-term test of at least 90 days, ideally spanning a heating season, placed in the lowest lived-in level. Short-term tests remain useful as screening tools when time pressure exists, such as a real estate transaction, but should be followed by confirmation.
Control the variables you can. Keep windows and exterior doors closed for at least 12 hours before and during a short-term test, run the HVAC normally, and avoid fans that vent to the exterior during the test window. Place the device 20 inches to 6 feet above the floor, at least 4 inches from walls, and away from drafts.
Retest after any structural change. A new furnace, a finished basement, a sump pump replacement, or foundation work can shift the annual average by 30% or more. A mitigation system should be retested within 30 days of installation and every two years thereafter, because the systems age and fan performance drifts.
Do not chase single readings. A continuous monitor showing 2 pCi/L at noon and 9 pCi/L at midnight is describing a normal pressure-driven system, not a crisis. The decision metric is the long-term average in the lowest occupied space.
What the data cannot tell you
Radon science has real limits that honest explainers should state. Exposure risk is estimated from occupational cohort studies of uranium miners extrapolated to residential levels, and the dose-response at low concentrations remains a subject of active epidemiological debate. The EPA's risk estimates—roughly 21,000 lung cancer deaths annually in the US attributable to radon—are model-based, not directly observed in residential populations. Year-to-year variability in any single home has not been characterized well enough for regulators to specify a precise confidence interval around a given test result.
What is well established is directional: radon is a Group 1 carcinogen, the second leading cause of lung cancer after smoking, and the risk rises with concentration and duration of exposure. The mechanism—alpha particle damage to bronchial epithelium—is not controversial. The measurement uncertainty is real, but it argues for longer, more careful testing, not for ignoring the number. If your home averages above 4 pCi/L over a heating season, mitigate. If it hovers between 2 and 4 pCi/L, confirm with a long-term test and weigh the case for a mitigation system, which typically costs 800 to 2,500 USD depending on the foundation type and the number of suction points in a US market. That price is a planning figure, not a quote, and varies by region and system design.
The core takeaway survives all the caveats: radon is a fluctuating signal, not a fixed property. Test long enough to average the noise, retest after changes, and treat the annual average in your lowest occupied floor as the number that governs the decision.