What is the most accurate quartz watch?

Two answers circulate. One says the most accurate quartz watch is an "atomic" or GPS watch, because it is always right. That answers a different question: a radio watch is right because it is corrected, not because it keeps good time on its own. The other answer names a chronometer-certified or high-frequency model, and mixes ratings that are not comparable, since one maker prints a yearly figure, another a daily one, and a third no number at all. We put the published figures from Citizen, Grand Seiko, Seiko, Longines, ETA, Breitling and Bulova on one scale, seconds per year without outside correction, with the COSC quartz limit beside them.

The ranking, maker by maker

Each row is the maker's own figure, or for Longines the figure reported at launch, converted to both units so the rows can be read against each other. The per-day column is the yearly figure divided by 365; the per-year column for daily ratings is the daily figure multiplied by 365.

Bulova's Precisionist does not appear because Bulova's own product page prints no number. It describes a "262kHz vibrational frequency", a "three-prong quartz crystal" and "unparalleled accuracy", and leaves it there. Figures for it that circulate elsewhere are not Bulova's, so we have not ranked it.

Why the 0100 is in a class of its own

Citizen describes the 0100 as telling "the precise time autonomously based on an internal mechanism with no reliance on time signals from radio towers or GPS satellites". Three things set it apart from the rest of the table, all from Citizen's own description:

• A different crystal. The 0100 uses an AT-cut crystal running at 8.4 MHz, 8,388,608 Hz exactly, which Citizen puts at 256 times the frequency of the tuning-fork crystal in an ordinary watch. Citizen chose the cut because it is "little affected by changes in temperature".
• Temperature checked every minute. The circuit measures temperature once a minute and adjusts the oscillator's frequency to match.
• More calibration points. Citizen doubled the number of temperature points at which each oscillator is tested before the watch is finished.

The catch is availability. Citizen's March 2019 release limited the references: 500 pieces of the AQ6021-51E in titanium at $7,400, 200 of the AQ6020-53X with a mother-of-pearl dial at the same price, and 100 of the AQ6010-06A in white gold at $16,800. Citizen's US site lists the AQ6021-51E as sold out. In practice the 0100 is now bought pre-owned; our page on Citizen covers the rest of the range.

The ±5 second tier is where most buyers should look

Four calibers from three makers share the next figure, and they get there by different routes. Citizen's A060 page says its crystal oscillators "were set aside for a six-month observation period before selecting only those that meet Citizen's demanding standards of accuracy". Grand Seiko says 9F accuracy is ensured "by the use of aged and individually selected quartz crystals", and reserves the ±5 rating for specially adjusted watches such as the SBGN001, a limited edition of 800. The Longines L288, a thermocompensated movement made exclusively for Longines by ETA and rated at ±5 seconds a year at its 2017 launch, adds gear position detection, which puts the hands back in place after a shock or a magnetic field knocks them out of step. Our page on how accurate a quartz watch is explains the temperature and aging effects that this selection and compensation are fighting.

For a buyer, ±5 seconds a year means a watch that is at most a few seconds out when you reset it at the clock changes. The current A060 reference we read, the AQ4100-57E, lists at $4,025. A Grand Seiko 9F lists at $2,600 to $3,400, as our Grand Seiko price guide sets out, though most of those are the standard ±10 grade rather than the specially adjusted ±5.

A COSC quartz chronometer is not the top tier

This is where most of the confusion comes from. COSC certifies quartz watches as chronometers, and a chronometer sounds like the best there is. The quartz test runs 13 days, in one position, at three temperatures and four humidity levels. The watch must average within ±0.07 seconds a day at 23 °C, stay within ±0.20 seconds a day at 8 °C and at 38 °C, and hold a rate stability of 0.05 seconds a day. COSC also requires a quartz chronometer to be thermocompensated.

Multiply that daily limit out and it comes to about ±25.6 seconds a year. That is about seven times tighter than an ordinary ±15 seconds a month watch, but looser than every yearly rating above it in the table. Breitling's B82 page states the figure plainly, "accuracy within ±0.07 seconds per day", and calls it "ten times more accurate than standard quartz". A passing watch has also proved itself for 13 days, not a year.

A movement can beat the COSC limit by far and carry the same certificate. ETA rates its PreciDrive movements at "+/- 10 seconds per year or 0.027 second per day" and says they have "no trouble meeting COSC criteria". So the word chronometer on a quartz dial tells you the watch is at least ±0.07 seconds a day. It does not tell you how much better. Our guide to accuracy and chronometer standards sets the quartz test beside the mechanical one, and our answer on what COSC certified means covers the certificate itself.

A higher frequency does not settle the ranking

High frequency is the selling point of several of these watches, but the table shows it is not the whole story. Grand Seiko's 9F manual gives its crystal frequency as 32,768 Hz, the frequency of an ordinary quartz watch, and it is rated at ±10 seconds a year. Bulova runs its Precisionist crystal at 262 kHz, eight times that frequency, and publishes no yearly figure at all. What decides the rating is how well the maker controls temperature drift and aging: thermocompensation, crystal selection and individual calibration. Frequency helps the 0100 because Citizen combined it with all three.

What the annual figure promises, and what it does not

A yearly rating describes a year of normal wear, not a cap on any single month. Grand Seiko's 9F manual is clear about it: the rating is ±10 seconds a year when the watch is worn about 8 hours a day at 5 to 35 °C, and the same manual warns that the monthly rate "may amount to approximately 2 seconds" depending on conditions. A 9F that is 2 seconds out after one month is within its specification, even though 2 seconds a month would add up to 24 in a year.

Seiko's service documentation for the 8F shows the other side of it. The technical guide gives the annual rate as "less than 20 seconds" across the normal temperature range, but the bench test a watchmaker runs is different: at 23 °C, with the case back closed, the normal rate is +0.020 to +0.140 seconds a day. Every value in that window is fast, and a watch at the top of it, if held at 23 °C all year, would gain about 51 seconds. The two numbers measure different things, a fixed-temperature bench rate and a yearly rate in use, so do not multiply a bench reading by 365 and call the watch out of specification.

Temperature is the other condition. Grand Seiko quotes its ratings at 5 to 35 °C and Citizen gives 5 to 40 °C as the 0100's optimal range. Outside that band every quartz watch drifts more, and our answer on cold weather and accuracy works out how much for an ordinary movement.

Where radio and GPS watches fit

A radio-controlled or GPS watch is the right answer if the question is "which watch shows the right time", and the wrong one if it is "which quartz movement keeps the best time". Between syncs it runs as an ordinary quartz watch, within ±15 seconds a month on the Casio and Citizen calibers in our page on whether an atomic watch will work where you live. A nightly sync means it never has more than a day's drift on it, about half a second. Miss the signal for a month and it can be 15 seconds out, where a 0100 would be well under a tenth of a second.

So if you live inside a transmitter's range, a radio watch that syncs is within about a second every morning for a fraction of the price. If you travel beyond every range or want a watch that depends on nothing outside it, buy a high-accuracy free-running caliber.

Which one to buy

• You want the best number there is: the Citizen 0100, pre-owned, at ±1 second a year.
• You want near-perfect timekeeping you can still buy new: a ±5 second caliber, the Citizen A060 or a Longines Conquest V.H.P., or a specially adjusted Grand Seiko 9F if you find one.
• You want a quartz watch built to be serviced for decades: a standard Grand Seiko 9F at ±10 seconds a year. Our page on whether Grand Seiko is worth the money weighs the price.

Mechanical watches are not in this race: their standards are measured in seconds a day, as our page on good accuracy for a mechanical watch sets out, and our quartz versus mechanical guide covers why people buy them anyway.

How to check the claim on your own watch

At ±5 seconds a year a watch drifts under half a second a month, below what you can read off a dial by eye, so a fair test takes six to twelve months. Set the watch to the second against an internet-synced clock, wear it normally, and read the difference at six and twelve months. The accuracy calculator turns the reading into a yearly rate. If a high-accuracy watch is minutes out after a few months, the problem is not the rating: suspect a weak cell or a fault, and our page on how long a watch battery lasts covers the first of those.

When this answer stops applying

• A new caliber is announced. This ranking is of the figures we read on 22 October 2026. A maker that publishes a tighter rating moves straight to the top.
• Your manual gives a different figure. The manual for your exact reference beats this table.
• The watch lives outside 5 to 35 °C. Every rating here assumes a temperature band; outside it, expect more drift.
• The watch is a smartwatch. Smartwatches take their time from the phone or the network, so this ranking does not apply to them.