What is a backflow test? What the tester measures, what passes, and what you have to do

The short answer: a backflow test is a field check of the mechanical assembly that stops water on your property from flowing back into the public water main. A certified tester connects a differential pressure gauge to the assembly's test cocks, closes and opens its valves in a set order, and reads whether each check valve holds and whether the relief valve or air inlet opens at the right pressure. The tester writes those readings on a report, and the report goes to your water utility. If a reading misses the published minimum, the assembly is repaired and tested again.

If a notice just arrived, the practical part is short: hire a tester on your utility's own list, let the tester file the report, and fix anything that fails. Our backflow tester directory lists the companies on each covered utility's list. The rest of this guide explains what the tester measures, what the numbers on your report mean, and why the rules differ between utilities.

What backflow is: two ways water runs backward

California's State Water Resources Control Board defines backflow as "an undesired or unintended reversal of flow of water and/or other liquids, gases, or other substances" into a public water system (Cross-Connection Control Policy Handbook, amended April 21, 2026). It happens in one of two ways, and the difference decides which device you need.

Backsiphonage: the supply side loses pressure

Backsiphonage "occurs when a non-potable source enters the drinking water supply due to negative (i.e., sub-atmospheric) distribution system pressure," the same handbook says. The causes it lists are mostly on the supply side: "main breaks, flushing, pump failure, or emergency firefighting water demand." Siphon action is stronger than people expect. The handbook notes it "has been shown to raise a liquid in a pipe such as water almost 34 feet."

Colorado State University Extension puts the lawn version plainly: a drop in pressure in the city main "can result in back-siphonage backflow of fertilizers, pesticides, manures and other contaminants through sprinkler heads and the irrigation piping system into the potable water supply" (CSU Extension, Home Sprinkler Systems: Backflow Prevention Devices).

The Texas Commission on Environmental Quality describes a real case on its cross-connection page: a worker had a garden hose at the bottom of a pesticide tank when crews opened a flush valve on the main nearby. The pressure dropped and pesticide flowed back through the hose into the house's water lines. Crews closed the valve before it reached the city's line, but the lines had to be flushed and customers in the area were warned not to drink the water until tests showed it was safe.

Backpressure: your side pushes harder than the main

Backpressure "occurs when the pressure from a non-potable source exceeds the pressure in the potable water distribution system" (California handbook). Houston Public Works lists typical sources: "pressurized tanks, reclaim systems outputting higher pressures than water supply, boiler systems" (Houston Cross Connection Control Program). On a lawn system, CSU Extension's example is a sprinkler pump or fertilizer injector: "the pressure exerted by the pump is greater than the pressure in the potable water system."

The split matters because two of the four testable devices only handle backsiphonage. Where backpressure is possible, they are the wrong device however well they test.

The four testable assemblies and where each is required

A backflow prevention assembly is, in the California handbook's definition, built so that "while in-line it can be maintained and its ability to prevent backflow, as designed, can be field tested." That last part is why your utility asks for a test: each of the four types has test cocks and shutoff valves so a tester can check it without removing it.

Sources: definitions and limits from the California handbook (sections 3.1.1 and 3.3.2); ASSE numbers as listed in Ohio Adm. Code 3745-95-06 and California handbook section 3.3.1. California's handbook also lets a water system substitute the detector version (DCDA, RPDA) where a DC or RP is required.

How the RP protects against both directions

The RP is the one the hardest cases get. California defines it as two check valves "with a hydraulically operating mechanically independent differential-pressure relief valve located between the check valves and below the upstream check valve." In plain terms: the space between the two checks stays at a lower pressure than the supply, because the first check holds back a few psi. If the first check leaks, or the supply pressure drops, that difference shrinks, and once it falls to the relief valve's opening point the valve opens and drains water from the space between the checks to the outside rather than letting it move back toward the main. USC's newsletters describe the same behavior from the tester's side: a relief valve that opens close to the first check's reading will discharge when line pressure fluctuates, and an RP whose relief valve fails to open while both checks leak is the case where backflow through the assembly becomes possible. The field test checks this point: the relief valve has to open while the gauge still shows at least 2.0 psid across the first check. California requires an RP's lowest point to sit at least 12 inches and, unless the water system approves otherwise, no more than 36 inches above grade; DCs installed or replaced after the handbook's adoption follow the same rule (CCCPH 3.3.2(b) and (c)).

Which one your property gets

The device follows the hazard, and the hazard is assessed by your water system. Four state rules show the pattern:

• California requires an air gap or RP for high hazard premises. Its Appendix D list includes medical facilities, dental offices with water-connected equipment, car washes, hotels and motels, gas stations, laboratories, mortuaries, and a "business park with a single meter serving multiple businesses" (CCCPH, section 3.2.2 and Appendix D).
• Washington requires an air gap, RPBA or RPDA for high health hazard premises, and accepts a DCVA for low hazards (WAC 246-290-490, Tables 12 and 13). Its high-hazard list includes "premises with separate irrigation systems using the purveyor's water supply and with chemical addition," such as parks, golf courses and cemeteries.
• Ohio ties the device to the degree of hazard: an air gap or RP for a "system or health hazard," and a double check is acceptable only for a "pollutional hazard." Its list of premises that need protection includes hospitals, car washes, and food or beverage processing plants. As an exception, a pressure vacuum breaker installed at the cross-connection to the irrigation system may be used as the "containment backflow preventer" for a water-only, residential-type irrigation system that is not subject to backpressure and has no pumps that could create it; the exception does not apply if an additive is used in the system, and "the supplier of water may determine other hazards exist that warrant a higher level of protection" (current text, effective October 10, 2022: Ohio Adm. Code 3745-95-05 and 3745-95-04).
• Texas leaves the selection table to 30 TAC 290.47(f) and requires "an air gap or backflow prevention assembly" at the meter wherever "an actual or potential contamination hazard exists," unless an adequate internal cross-connection control program is in place (30 TAC 290.44(h)). The City of Round Rock, for example, treats hospitals, restaurants and industries as typical high hazard uses and residential irrigation as a typical low hazard use (Round Rock backflow program).

At a house, common triggers include a lawn sprinkler system, a pool with auto-fill, a fire sprinkler system or a private well. Marin Water's examples of hazards that need protection include irrigation systems, hose bibs, boilers, wells, ponds, pools, spas and rainwater collection systems (Marin Water, Backflow prevention). A well is a common surprise: EBMUD requires customers with private wells to test a backflow preventer every year, and its FAQ says state regulations require the annual test even if the well only feeds irrigation (EBMUD, Backflow prevention). If you kept a well after connecting to city water, our guide to well water vs. city water covers that setup.

The atmospheric vacuum breaker (AVB) on some sprinkler zones has no test cocks. Washington has AVBs on irrigation systems inspected at installation, after a backflow incident and after repair instead of tested yearly (WAC 246-290-490(7)(c)).

What the tester's gauge actually measures

A backflow test kit is a differential pressure gauge. It does not tell the tester how much pressure is in your line. It has a high-side hose and a low-side hose, and it shows the difference between the two points they are connected to, in pounds per square inch differential (psid). Hook the high side upstream of a check valve and the low side just downstream, close the right valves, and the needle shows how much pressure that check valve is holding back. A USC newsletter describes it exactly this way for the second check of a double check: the gauge reading "gives us the differential pressure across the No. 2 check valve" (USC FCCCHR, Cross Talk, Spring 2001).

The gauge itself is regulated, because a test is only as good as the instrument:

• Texas requires gauges to "be tested for accuracy annually" under the USC manual or AWWA Manual M14, and the gauge serial number goes on the report (30 TAC 290.44(h)(4)).
• TCEQ's form instructions add that a gauge used on potable-water assemblies "must not be used to test non-potable backflow prevention assemblies" (Form TCEQ-20700).
• Round Rock asks testers for proof that their equipment holds calibration to "plus or minus 0.2 psid accuracy."

The state rules in this guide point to the same procedure manual: the Manual of Cross-Connection Control, Tenth Edition, published by the University of Southern California Foundation for Cross-Connection Control and Hydraulic Research (USC FCCCHR). California's handbook names it for assembly approval and for the field test procedures testers are examined on ("or equivalent"), Washington accepts test procedures "such as those specified in the most recently published edition of the USC Manual," and TCEQ says a Texas tester's license "is based on the testing procedures described in the University of Southern California's 10th edition manual."

The annual field test, device by device

Each device gets a short sequence of readings. The RP and double check minimums below match the values in USC's own documents, and one Florida utility, the City of Wilton Manors, lists all of them on a one-page owner fact sheet, which is the source of the chart (Backflow Assembly Tests Fact Sheet, updated 04-09-24).

Reduced pressure principle (RP): three readings

1. Relief valve opening point. The tester lowers the pressure difference across the first check and notes the reading at which the relief valve starts to discharge. Wilton Manors: "The relief valve should open before the gauge drops below 2.0 psi." If it opens at 2.0 or higher, that reading passes; if it stays shut until 1.8, it fails.
2. Check valve No. 1. Under the USC Tenth Edition, the first check "must be above the relief valve opening point and ≥ 5.0 psid" (USC FCCCHR, Field Test Procedures 9th vs. 10th Edition).
3. Check valve No. 2. Tested against backpressure and recorded as closed tight or leaked, not as a number. Texas' form says it outright: a numeric second-check reading is "required for DCVA only."

Why a separate number for the first check? The Ninth Edition said the first check "should be 3.0 psid greater than the relief valve opening point." In 2001, USC explained that this 3-psi buffer "does not affect the ability of the assembly to prevent backflow but is there to minimize the nuisance of water discharging intermittently when the line pressure fluctuates," and said the Tenth Edition would drop it and set the first-check minimum at 5.0 psid. USC's stated reason for the change: "To provide a required minimum value for the first check valve." If an older form in your files mentions a 3.0 buffer, that is why.

Double check valve assembly (DCVA): two readings

Each check valve must hold at least 1.0 psid. A 2001 USC newsletter on the double check test shows why readings near zero are a problem for the procedure itself: if check No. 1 reads below 1.0 psid and the No. 1 shutoff valve leaks, the tester has to stop and repair check No. 1 before testing check No. 2. Technically the second check could still be evaluated with the first holding 0.1 to 0.9 psid, "But now, the absolute accuracy of a tester's gage at the low end of the scale would be critical," so USC's committee set that cut-off at 1.0 or greater, which gives "a reasonable safety factor."

Pressure vacuum breaker (PVB) and SVB: two readings

For a PVB, Wilton Manors lists: "The air inlet valve must open before the gauge drops below 1.0 psi" and "The check valve must hold a minimum of 1.0 psi." The report also asks whether the air inlet opened fully. An SVB is tested on the same two parts, check valve and air inlet. The USC Tenth Edition changed the order for the SVB so that the check valve reading comes first and the air inlet second, "To simplify the field test procedure."

How to read your test report

The report is where the gauge readings end up, and you can check it yourself. In Texas the format is set by Form TCEQ-20700; other states and utilities use their own forms with the same boxes. The rows below use the Wilton Manors minimums and the USC Tenth Edition rule for the RP first check.

Other boxes worth checking: the assembly's serial number and size (they should match the device in your yard), the gauge serial number and the date it was last tested for accuracy, and the tester's certification or license number. In Texas the tester must sign and date the original, which "must be submitted to the public water supplier for recordkeeping purposes" (30 TAC 290.44(h)(4)).

Is backflow testing really annual? It depends on your utility

"Annual backflow testing" is the rule at most of the utilities quoted in this guide. It is not universal, though. Texas' rule requires a test when any required assembly is installed, but the yearly retest applies to assemblies "installed to provide protection against health hazards." That leaves room for a utility such as Round Rock to set a longer interval for low-hazard devices.

Two details in that table are easy to miss. California counts draining the assembly for winter as a reason to retest, so a sprinkler assembly that was drained in fall may need a test when it goes back into service. And Texas' "12 months" is measured from the last test, not from the calendar year.

Who may test, and why your utility keeps its own list

You cannot do the official test yourself, even with your own gauge, unless you hold the credential. Every state rule above requires a qualified tester:

• Texas: testers must complete an approved course, "pass an examination administered by the executive director, and hold a current license." Fire lines are separate: testers "may test and repair assemblies on firelines only if they are permanently employed by an Approved Fireline Contractor" (30 TAC 290.44(h)(4)).
• California: testing must be done by a person certified by an organization the State Water Board recognizes. That certification requires a proctored written exam of "no less than 100 questions for initial certification" (CCCPH 3.4.1).
• Ohio: tests are performed "by a person approved by the supplier of water and acceptable to the director," and they are "at the expense of the water consumer" (3745-95-06).
• Washington: a backflow assembly tester (BAT) does the testing (WAC 246-290-490).

Many utilities add their own layer, which is where the lists come from. EBMUD testers must pass "a written and hands-on examination administered by East Bay Municipal Utility District Backflow Dept. staff," and the district says a test "must be performed by a tester on EBMUD's Approved Backflow Testers List." Marin Water schedules hands-on demonstrations in April and October and warns that if a test is done by someone not on its list, it "cannot accept the test results." Houston requires a TCEQ license plus city approval, and publishes a testing company list that is "updated monthly." A state-licensed tester may still not count in your city, so check the list for your utility.

The lists we copy, with their version dates, include Houston, Round Rock, EBMUD (Oakland), Marin Water, Sacramento and Columbus. A listing is not an endorsement by the utility or by us.

What you have to do after a notice

1. Read the notice for three things: the due date, the device or account it refers to, and any confirmation number. Utilities give different lead times: Round Rock sends notices 30 days before the due date and asks you to give the tester the Customer Confirmation Number printed on it; Marin Water mails postcards 45 days ahead. When EBMUD requires a new assembly, it gives 30 days to install and test it and asks for the test form within 30 days of notification.
2. Hire a tester on your utility's list. Find your utility in the directory. For a fire line in Texas, the tester must work for an approved fireline contractor; Houston marks those companies on its list. EBMUD suggests that "obtaining three bids ensures competitive pricing." What a test usually costs is in our backflow testing cost guide.
3. Make the assembly reachable. California requires each assembly to be "accessible for field testing, inspection, and maintenance." Clear the box lid, plants or stored items. If you winterized the system, tell the tester; Salt Lake City's guide recommends draining the assembly by opening the test cocks and turning the shutoff valves to a 45-degree angle for winter, so valves may not be where the tester expects (SLC winterizing guide).
4. Expect a short visit and a brief shutoff. The test closes the assembly's shutoff valves, so water through it stops while the tester works. Wilton Manors says the field test "takes about ten minutes," and that an initial test, a cleaning, new rubber parts and a final test can usually be done in one hour for assemblies 2 inches and smaller.
5. Make sure the report is filed. In most programs the tester files it. Round Rock and Houston require testers to submit reports through an online system the city specifies; in Texas the signed original goes to the water supplier. Ask for a copy. Ohio puts record-keeping on the owner: "Records of such inspections, tests, repairs and overhauls shall be kept by the consumer and made available to the supplier of water."
6. If it fails, repair and retest on the deadline. California gives 30 days from notification of a failure, with extensions only if the water system's plan allows them. Round Rock also gives 30 days to repair and retest. Houston is blunt: a failed test leaves you out of compliance "until a passing test report is submitted." Repairs are often small. Wilton Manors notes "Often the assembly only needs to be cleaned," says these assemblies are designed to last from 20 to 50 years, and advises against replacing one unless parts are no longer available or the body is damaged.

What testers notice besides the readings

A tester also looks at the installation, because a perfect reading on the wrong device or in the wrong place protects nothing. California sets these dimensions:

Texas' report asks whether the assembly is "installed in accordance with manufacturer recommendations and/or local codes." California also tells water systems to require testers to report within 24 hours if they see a backflow incident or an unprotected cross-connection during a test (CCCPH 3.3.3(f)).

Homes, restaurants and managed buildings: what is different

Homes with lawn irrigation most often have a PVB or an RP on the sprinkler supply, according to CSU Extension. Adding a fertilizer injector or a pump can change the rating: CSU says chemical injection "represents a higher degree of hazard and requires the use of a device such as the reduced pressure principle backflow assembly."

Restaurants and commercial buildings show up on more high hazard lists. Round Rock names restaurants among its typical high hazard uses, which means annual testing there. California lets water systems test more often for high-hazard premises and for assemblies "at increased risk of testing failure." Houston adds that a device at the meter is "additional protection" and does not replace the plumbing code's protection on internal hazards.

Property managers often have several assemblies on one site, each with its own serial number and due date, so keep a list that matches the utility's records. California treats a business park on a single meter as high hazard. If an assembly was removed or replaced, Houston wants a removal form from your plumber, engineer or testing company.

Next step

Find your water utility on the backflow tester directory, open its list, and book a tester on it before the date on your notice. Texas owners can start at the Texas lists, California owners at the California lists. When the report comes back, check the readings against the table above. If your utility is not in the directory yet, the number on the notice will tell you where its list is.

Questions people ask