How to Request a Water Meter Accuracy Test in California (and Why It Rarely Wins)

Published by Clément Lacaille, Tech-Bharat · Updated September 8, 2026 · How we research

Every plumber in California has had this conversation. A homeowner opens a water bill that's two or three times normal, checks everything they can think of, finds nothing running, and lands on the obvious suspect: the meter must be wrong. It's a completely reasonable thought, and California utilities have a formal process for testing it — one that is cheaper and more accessible than most people realize. But before you file that request, you should know something that almost nobody tells you up front: when mechanical water meters drift with age, they overwhelmingly drift slow, not fast. Understanding that changes both how you approach the dispute and what you should be looking at instead.

First, actually rule out the leak — because it usually is one

A meter test is a slow, low-yield move if you haven't finished the free diagnostic work first. Shut off every fixture and appliance in the house, then watch the meter's leak indicator (the small triangle, star, or dial that spins on the slightest flow) for fifteen minutes. If it moves at all, you have flow somewhere, and the meter is doing its job. Our guide on how to read your water meter and check for a hidden leak walks through that test step by step, including the trick of closing the house shutoff valve to separate a leak inside the house from one in the service line between the meter and the building.

Then account for the things that produce a big bill without a spinning indicator: a toilet flapper that leaks intermittently and reseats between cycles, an irrigation controller that got reprogrammed or ran through a hot spell, a water softener stuck in regeneration, a pool auto-fill, a recirculation loop, a longer billing period than usual, or a rate tier you crossed. The hidden causes behind a high water bill guide covers the ones that hide best. Also check the number of days in the billing cycle and whether the read was estimated rather than actual — an estimated read followed by a true-up read produces a spike that looks exactly like a leak but is really a catch-up.

The uncomfortable part: worn meters almost always under-register

This is the honest core of the subject, and it's worth saying plainly. Mechanical water meters — the nutating-disc and piston meters under most older California meter boxes — measure by physically moving a part every time water passes. That part wears. Bearings loosen, chambers score, grit erodes clearances. The consistent result of that wear is that water slips past unmeasured, especially at low flow rates, so the meter registers less than actually passed. It does not spontaneously start counting water that never flowed.

The research is quite consistent on this. An empirical study of residential meter degradation by Fourie, Marnewick and Joseph, published in Water SA in 2020, tested meters pulled from service and found errors that were negative — under-registration — across the board, with velocity-type meters losing roughly 1.13% of accuracy per 1,000 kilolitres of accumulated throughput, and accumulated volume proving a better predictor of a meter's error than its age. That direction of error is exactly why utilities run meter replacement programs at all: an old meter is a revenue problem for them, not for you.

What about simply arguing that your meter is twenty years old and therefore suspect? Age alone is weaker evidence than it sounds. A 2018 study by Arregui and colleagues in the journal Water tested two types of single-jet residential meters under comparable conditions and found significantly different degradation curves between them, concluding that accuracy loss depends on the specific design and the real working conditions in the field rather than following one predictable timeline (see the study). Two meters installed the same year on the same street can be in very different metrological shape.

None of this means a fast meter is impossible. Meters do occasionally fail in strange ways — a damaged register, a stuck odometer wheel, air passing through the measuring chamber after a main break or a service restoration, or, more commonly than an actual fast meter, a meter that is simply assigned to the wrong account. If your neighbor's meter and yours got crossed in the billing system, the numbers on your bill are perfectly accurate for somebody else's house. That one is worth raising explicitly when you call.

What "accurate" officially means

The benchmark is set by AWWA standards. For new displacement-type residential meters, AWWA's C700 standard calls for registration in a band of roughly 98.5% to 101.5% of true volume at normal and high flow rates, with a wider allowance at the specified low test flow. In California, the CPUC's General Order 103-A sets the accuracy requirements and test procedures that regulated water utilities must follow, and municipal agencies generally write similar tolerances into their own municipal codes. The important consequence is that the test is pass/fail against a tolerance, not a search for a satisfying explanation — a meter that reads 1% fast is a passing meter, and a passing meter ends the dispute.

Requesting a test from a CPUC-regulated utility

If your water comes from an investor-owned utility — California Water Service, Golden State Water, San Jose Water, San Gabriel Valley Water, Suburban Water Systems, California-American Water, Great Oaks Water, Liberty Utilities and the other CPUC-regulated companies — your rights are written into a published tariff, and you can read it yourself. Look for the company's Rule No. 18, titled "Meter Tests and Adjustment of Bills for Meter Error." The terms vary slightly by company, but the California Water Service version is representative and worth knowing in detail:

The utility must proceed to test the meter within one week of a customer request. There is no charge for a customer-requested test, except where the customer asks within six months of the meter's installation or more often than once a year — in those cases the customer pays the cost. The meter is tested "in the condition as found," before any adjustment or cleaning, which matters because a meter cleaned first would test better than it ran. Testing is done at the three flow rates specified in General Order 103 plus twice the minimum test flow, and the reported average error is the algebraic average of the errors at the three highest test flows. You have the right to require that the test be conducted in your presence or that of your representative, and the utility must furnish you a written report of the results within 15 days of receiving them.

The remedy is also spelled out. If the meter tests more than 2% fast, the utility refunds the overcharge based on corrected readings for the period the meter was in use, capped at six months. And here is the part people don't expect: the rule runs both directions. If a residential meter tests more than 5% slow, the utility may back-bill you for the undercharge, capped at three months. A non-registering meter can also be back-billed on an estimated basis for up to three months. In practice utilities don't always pursue small undercharges, but the authority is there in the tariff, and given that under-registration is the far likelier finding, it is a real if modest downside to weigh.

Requesting a test from a city or water district

If you're served by a municipal utility or a special district — most of California's population is — the CPUC has no jurisdiction and the rules live in the local municipal code or the district's rate schedule. The common structure there is a deposit rather than a free test: you pay a fee up front to cover the cost of pulling and bench-testing the meter, and the fee is refunded only if the meter fails tolerance. San Luis Obispo's municipal code, for example, requires a deposit in an amount set by council resolution, returns it if the meter registers more than 2% fast, and refunds the overcharge on corrected readings for up to a year; it also entitles the customer to at least five days' notice of when and where the test will happen. San Clemente uses a 3% threshold and states outright that the test fee is forfeited to the City if the meter comes back within tolerance. Deposit amounts vary widely by agency and are typically modest — the point of the fee is to deter casual requests, not to fund the utility.

Ask your agency three questions before you commit: what the fee is and whether it is refundable, what accuracy threshold triggers a refund and a bill adjustment, and how far back the adjustment goes if you win. Ask for the answer in writing or a link to the code section, and ask whether they will do a simple on-site field test first — many agencies will run a known volume through the meter at the box, at no charge, which can settle an obvious failure without pulling the meter at all.

What a new AMI or smart meter changes

If your agency recently swapped your old mechanical meter for an AMI or solid-state smart meter, the situation is genuinely different, and the difference cuts against the intuitive complaint. Ultrasonic and other solid-state meters have no moving measuring parts to wear, and they register very low flows that a worn mechanical meter would have let slip by uncounted. So a household that changes nothing at all can see its billed consumption rise after a meter swap — not because the new meter is wrong, but because the old one had quietly stopped counting a slice of the water. That is an unwelcome answer, but it's the correct one often enough that it should be your first hypothesis after a recent meter change.

The genuine upside is the data. An AMI meter typically records hourly or sub-hourly interval reads, and most California agencies expose that data in a customer portal. That interval data is a far better diagnostic than a meter test: a flat nonzero baseline every hour of the night is a continuous leak, a repeating identical block at 4 a.m. is irrigation, and a single enormous spike on one afternoon is a hose left running or a line break. Our guide on what to do when your smart water meter sends a leak alert covers how to read those alerts and what they actually detect. Pull the interval data before you request a test — it will usually tell you what the test would not.

When the test comes back "accurate"

Most tests come back within tolerance. That isn't a dead end; it's a redirection, because it tells you the water genuinely went through your service line. At that point the productive moves are:

Ask the utility to confirm the meter serial number on the tested meter matches the one on your account, and to verify the meter reads on the disputed bills against the current physical read — transcription errors and crossed accounts survive a meter test perfectly well because the meter itself was never the problem. Ask whether the bill was based on an estimated read. Ask whether a rate change or a tier restructure landed in the disputed period; our guide to 2026 California water and sewer rate increases explains why the same volume of water costs meaningfully more than it did a couple of years ago, which by itself explains a share of "my bill doubled" complaints.

Then bring in a plumber for a pressure test of the service line and a leak-detection sweep, because a leak that runs underground between the meter and the house, or under a slab, can consume an enormous volume without ever showing at the surface in California's dry season — the ground simply takes it. If a leak is found and repaired, you may qualify for a one-time courtesy adjustment: see our guide to the water bill leak adjustment credit in California, which is a separate and much more commonly granted remedy than a meter-error correction. Most agencies require proof of repair — keep the invoice.

One structural thing worth checking while you're at it: if your meter is undersized or oversized for the property, that affects measurement quality at the extremes of flow, and it's worth understanding before you assume the meter is defective. Our guide on water meter upsizing costs in California covers when a size change is actually justified.

The bottom line

Requesting a meter accuracy test is cheap, it's your right, and there's a real process behind it — for CPUC-regulated customers it's usually free once a year under Rule 18, and for municipal customers it's a modest refundable deposit. Do it if you've genuinely exhausted the alternatives, and do it with your eyes open: the physics of meter wear runs against you, and a slow meter can theoretically be back-billed. In the overwhelming majority of high-bill cases, the water really did flow, and the money is in finding where. If you've ruled out the obvious and want a professional to pressure-test the service line and hunt for what the meter is honestly reporting, request a free quote and a licensed local plumber can locate it before another billing cycle goes by.

Keep reading

• CSLB License Check
• Emergency Plumber Costs in California
• Water Heater Lifespan
• Tank vs Tankless Water Heater in California
• California plumbing cost guides