For a real estate transaction, a 2-to-7-day short-term test under closed-house conditions. For deciding about your own home, a long-term test of 90 days or more, because it captures the annual average the EPA guidance is written against.
More detail
The two tests answer different questions. A short-term test measures a snapshot of 2 to 7 days under closed-house conditions, meaning windows and exterior doors stay shut except for normal entry. It is fast and it is the standard for a purchase contract, where the measurement protocol is ANSI/AARST MAH-2023 with chain-of-custody documentation for the buyer, seller, and lender file. A long-term test runs 90 days or more and produces something closer to your actual annual average exposure. That matters because radon levels swing substantially with weather, barometric pressure, and season. The same house frequently reads higher in winter, when the stack effect of a heated house pulls harder on the soil beneath it and windows stay closed. That seasonal swing is why two tests on the same house can disagree without either being wrong. A summer kit reading 3.2 and a winter kit reading 5.8 are both accurate measurements of different conditions. If two results disagree and a decision hangs on it, the long-term result is the better basis, and a continuous radon monitor placed by a measurement professional resolves it faster. Hardware store charcoal kits are reasonably accurate when used correctly, and most reported inaccuracy traces to protocol rather than the kit: placed too close to an exterior wall or a draft, disturbed mid-test, left too long before mailing, or run with a window open. For a transaction, use a professional test regardless, because the chain of custody is part of what the buyer and lender are paying for. Seasonal swing matters more here than national guidance implies. Central Ohio sits on flat-lying glacial till with dense clay subsoils that drain slowly and perch water above the till rather than letting it through. Homes in this market commonly read higher between November and March, so a summer short-term result of 3.0 is not evidence the house sits under the action level across the year.