Five things decide whether a pressure test counts: how hard you pump it, how long you hold it, what you pump into it, what you read it with, and what you hand over afterwards. Here is what the model codes ask for each one — and where your jurisdiction is likely to disagree.
The test pressure is one and a half times the pressure the piping will actually carry, and never less than 3 psi. On a standard low-pressure residential system that lands on the floor value: 3 psi. On a 2 lb system, 1.5 × 2 is also 3 psi. It is only above roughly 2 psi of working pressure that the multiplier starts to bite — a 5 psi system tests at 7.5 psi. IFGC 2021 §406.4.1 · IRC §G2417.4
With CSST, there is a ceiling as well as a floor: never test above the tubing's rated pressure.
This is the number most often quoted wrongly, including by search engines. The IFGC rule is not "ten minutes" — it is half an hour for each 500 cubic feet of pipe volume, or fraction thereof. Ten minutes is the exception: it applies when the system holds less than 10 cubic feet, or when it is in a single-family dwelling. No test has to run longer than 24 hours. IFGC 2021 §406.4.2 · IRC §G2417.4
In practice that means a house is a ten-minute test and a commercial run is not. A hundred feet of ¾-inch pipe holds about 0.4 cubic feet, so a dwelling never comes near the threshold — but five hundred feet of 2-inch pipe is roughly 12 cubic feet, and that is a half-hour hold before anyone argues about it.
| System | Minimum hold |
|---|---|
| Single-family dwelling | 10 minNamed exception, whatever the volume |
| Under 10 ft³ of pipe | 10 minThe other named exception |
| 10 – 500 ft³ | 30 min½ hour for the first 500 ft³ or part of it |
| 501 – 1 000 ft³ | 60 min½ hour for each further 500 ft³ or part of it |
| Over 24 000 ft³ | 24 hThe test is not required to run longer |
The calculator works the volume out from pipe size and length →
Jurisdictions on the Uniform Plumbing Code work to a different figure again: not less than 10 psi held for 15 minutes with no perceptible drop, and for welded piping or systems over 14 inches of water column, not less than 60 psi for not less than 30 minutes. UPC §1213.3
If your permit or your inspector names a duration, that duration wins — and local amendments are common. Nashville's residential amendment, for instance, asks for 10 psi held for twenty minutes; Hawaii Gas publishes 10 psig for 15 minutes; Massachusetts requires systems over a million BTU to be tested at ten times the working pressure. None of those are in the model code.
Air, nitrogen, carbon dioxide or an inert gas. Oxygen is named outright as forbidden; the fuel gas itself is simply not on the permitted list. IFGC 2021 §406.2 · IRC §G2417.2 · UPC §1213.1.6
Oxygen is the dangerous mistake here. Compressed oxygen in piping with any oil or grease residue is an ignition hazard, which is why every rule set names it explicitly.
Read the test with a manometer, or with a dial gauge whose top of scale is no more than five times the test pressure. For a 3 psi test that means a gauge reading up to 15 psi — not the 60 psi gauge in the bottom of the van, because a 0.5 psi loss on a 60 psi dial is invisible. IFGC §406.4
Many jurisdictions tighten this further: 0.1 psi increments, a range no more than twice the test pressure, or a requirement that the reading sits in the middle half of the dial. Check before the inspection, not during it.
Whether a written test report is accepted instead of a witnessed test is the inspector's call. That permission lives in the administrative chapter of whichever code your jurisdiction adopted, not in the gas chapter — so the honest answer is that it is at the discretion of the Authority Having Jurisdiction, and you ask before the inspection.
That is exactly why the record matters. A report that names the address and section tested, the gas and the medium, the test pressure and the hold, the gauge and its range, the time the hold started and ended, photographs of the gauge at both ends, and the technician attesting to the result is hard to argue with. A pressure reading written on the back of a permit is not.
Air that goes into the pipe warm reads high, and cools while you wait. On a cold morning a system charged straight from a compressor can lose a readable amount of pressure with nothing wrong with it. Charge it, let it settle, then start the hold from the settled reading — and note the temperature swing if you have to explain the drop.
The other usual suspects: the test gauge itself, a valve left open into an appliance, thread sealant that has not cured, and a test tee that seals against the gauge but not against the plug.
GasTest Log times the hold on the wall clock, keeps a start and an end photo of the gauge, and exports a one-page certificate carrying these same code citations. The app never measures pressure and never decides the result — you read the gauge and attest to it, which is exactly what your licence is for.