Timeless TicksWatch knowledge, checked Search Menu

Answer

Why is my new watch losing time?

Short answer

Most new watches that lose time are not faulty. The usual causes are an automatic that never reaches full wind, a rate that is inside the maker's published tolerance but outside your expectations, and magnetism picked up in the first weeks. Measure the mean daily rate over seven days before doing anything, then compare it with the figure the manufacturer actually publishes.

On this page (7 sections)
  1. What "new" actually changes
  2. The automatic that never reaches full wind
  3. The seven-day measurement protocol
  4. The tolerances a new watch is actually held to
  5. Magnetism in the first weeks
  6. When it is a genuine fault
  7. Making a claim that works

Four different situations produce the same complaint, and they need different responses. A watch settling in after assembly, an automatic that is simply not getting enough wrist time, a movement that has picked up a magnetic field, and a genuine fault all read on your phone as minus a few seconds. Before you contact anybody, establish which one you have.

The order below is deliberate. The first two explanations are free and account for most cases. The third takes a watchmaker ten seconds. Only the fourth is a warranty matter.

What "new" actually changes

A movement is regulated at the factory, then cased, then packed, and it may have sat in a warehouse or a display case for months before you opened the box. Three things follow from that.

Fresh lubricant is still distributing across the pivots and the escapement during the first weeks of running, and a new mainspring takes a little time to settle. Both can move a rate by a few seconds a day. That is the real content of the phrase "running in", and it is why most watchmakers will not regulate a new watch until it has run for a month.

The second change is that the watch has been static and is now on a moving arm, at your skin temperature, in your resting position at night. Factory regulation is done to an average across positions. Your particular mix of positions may sit at one end of that average.

The third is the least obvious: a watch that has been sitting unworn is at zero wind when you first strap it on, and it takes a full day of wear or a proper hand-wind to get near full. Rate measured during that first day is measured at the worst part of the power curve.

The automatic that never reaches full wind

This is the most common single cause and it is not a fault. A mechanical watch keeps its best time near full wind, and an automatic depends on your arm to stay there. Desk work, short wear hours or a loose bracelet leave it running on the weak end of the mainspring, where losing five or ten seconds a day is ordinary. If your new watch also stops overnight, that is the same problem showing its other symptom, and our answer on watches that stop overnight deals with it directly.

Rule it out before anything else. Wind the watch to full by hand, roughly 30 to 40 turns of the crown from a stop, then keep winding it fully each morning through the test week. You cannot overwind an automatic. If the loss shrinks or disappears, the movement is healthy and the problem was the power reserve. If it does not, you have removed the biggest variable from the measurement. The broader list of causes for a watch that has always run slow is in why is my watch running slow.

The seven-day measurement protocol

One morning's reading is not a rate. Position, temperature and state of wind move a mechanical watch by several seconds from one day to the next, so a single figure tells you almost nothing. Seven consecutive days does tell you something.

  1. Set it against a reference. Use an internet-synced clock, not another watch and not the microwave. Set the seconds precisely with the crown pulled out so the balance is stopped, if your movement hacks.
  2. Wind it fully each morning, at the same time, before you read it.
  3. Read the deviation at the same time every day, to the second, and write it down as a running total rather than as a daily difference.
  4. Note the overnight resting position each night, and keep it the same all week. Dial up and crown down can differ by ten seconds a day on the same watch.
  5. Wear it normally otherwise. The number you want is the rate you actually get, not a laboratory result.
  6. Do not reset it mid-test. Resetting destroys the measurement.
  7. Divide the total deviation by seven to get the mean daily rate. The accuracy calculator does the arithmetic and shows the spread.

Our note on testing accuracy at home covers the positional test and what a timing app or a timegrapher adds, including amplitude, which is the reading that separates an adjustment from a repair.

Accuracy standards compared on a seconds per day scaleA number line from minus 30 to plus 40 seconds a day. METAS Master Chronometer spans 0 to plus 5. COSC chronometer spans minus 4 to plus 6. A good non-certified automatic spans minus 10 to plus 20. An entry level automatic spans minus 20 to plus 40. Quartz sits far off the chart at about half a second a day.How accurate is accurate?Seconds gained or lost per day, drawn to scale-30-20-100+10+20+30+40perfectMETAS Master Chronometer+0 to +5COSC chronometer-4 to +6Good non-certified automatic-10 to +20Entry level automatic, as sold-20 to +40A quartz watch sits far off this chart: roughly 15 seconds a month, about 0.5 seconds a day.
Accuracy bands. Certification buys a narrower window, not a different kind of timekeeping. Any mechanical watch is beaten by a $30 quartz.

The tolerances a new watch is actually held to

Now compare your figure with the one the maker publishes for your caliber, not with your expectations. The bands are wider than most new owners assume.

Published rate specifications, in seconds per day unless stated
SpecificationPublished toleranceHow it is measured
Seiko 4R calibersplus 45 to minus 35Worn on the wrist, 5 to 35 degrees Celsius
Seiko Instruments NH35Aminus 20 to plus 40Static, movement specification
Seiko 6R calibersplus 25 to minus 15Worn on the wrist, 5 to 35 degrees Celsius
COSC chronometerminus 4 to plus 6 mean daily rateMovement, before casing, across positions and temperatures
Oris Calibre 400minus 3 to plus 5Brand specification
Tudor, COSC modelsminus 2 to plus 4Cased watch, brand specification
Rolex Superlative Chronometerminus 2 to plus 2 averageCased watch, brand specification
METAS Master Chronometer0 to plus 5Cased watch, independent certification

Two things to take from the table. A watch losing eight seconds a day on an affordable Japanese automatic is comfortably inside specification and nothing is wrong with it. And a COSC figure is a movement result measured before the movement went into a case, so a certified watch that runs a second or two outside the band on your wrist has not necessarily failed anything. Our guide to accuracy and chronometer standards explains what each certificate covers, and what COSC certification means covers that one in detail. If you want the plain version of what to expect, see what counts as good accuracy.

Magnetism in the first weeks

A new watch is handled more than an old one: it goes on and off the wrist, gets photographed on a laptop lid, sits on a charging pad or beside a phone with a magnetic mount. Magnetism most often makes a watch gain, sometimes by minutes a day, but a magnetized rotor or barrel adds drag instead and can make it lose.

It costs nothing to rule out. A watchmaker passes the watch over a demagnetizer for a few seconds, and many will do it while you wait for free. Do it before you start a warranty conversation, because a magnetized watch returned under warranty comes back demagnetized with a note saying no fault found. What magnets do to a watch covers the sources and the test.

When it is a genuine fault

Some patterns are not normal on a new watch at any price:

  • A mean daily rate outside the manufacturer's published band after seven days of full wind, with the watch demagnetized.
  • A loss that grows week by week rather than staying steady.
  • A watch that runs while worn and stops when rested, which points to low amplitude.
  • A rate that changed suddenly after a knock or a drop.
  • Any moisture, fogging or debris visible under the crystal, which is a separate and more urgent problem.
  • A power reserve well short of the published figure, measured by winding fully and leaving the watch until it stops.

Assembly faults do reach customers. A reversing wheel with the wrong lubricant, a hairspring touching the balance arms, or a rotor bearing that binds all present as poor timekeeping on a watch fresh from the box, and all of them are warranty work rather than something to live with.

Making a claim that works

Take the seven-day log with you, or attach it. A dated table of readings against a stated reference, with the resting position noted and a line saying the watch was fully wound each morning and demagnetized on a given date, is a different conversation from "it loses time". It shows a mean daily rate, it shows consistency, and it pre-empts the two easy dismissals.

Ask for regulation to the published tolerance, in those words. Inside the warranty period, which commonly runs two to five years, adjusting a watch that does not meet its own specification is normally free. Outside it, regulation is a small paid job, typically in the region of $60 to $150 in the US, before sales tax, and more or less than that elsewhere depending on local labor rates. Expect the watch to be away for several weeks, and ask for the timing result on return. The manufacturer's warranty is the contract that travels with the watch wherever you live, and whatever rights you have against the seller sit alongside it rather than replacing it: in the US those come from the retailer's own returns policy and from state law, while in many other countries a statutory guarantee applies whose length is set by national law, so check the rule where you bought. Our guide to warranties and consumer rights sets out how the two interact and what a distance purchase adds.

One last piece of perspective before you send anything back. If the watch is inside its published band and you simply want better than that, say so plainly: you are asking for an adjustment, not reporting a defect. Many watchmakers will regulate a healthy movement to within a couple of seconds a day for a modest fee, and on a well-made watch that adjustment usually holds.

Common follow-up questions

How long is the running-in period on a new mechanical watch?

There is no formal one. Fresh lubricant redistributing and a new mainspring settling can move a rate by a few seconds a day over the first few weeks, and most watchmakers will suggest waiting a month before regulating a new watch. That is the honest size of the effect. A watch losing thirty seconds a day out of the box is not settling in, and waiting will not fix it.

The dealer says a mechanical watch is not meant to be accurate. Is that true?

Half true, and often used to end a conversation. Mechanical watches are less accurate than quartz by two orders of magnitude, which is a fact about the technology. But every caliber has a published tolerance, and yours either meets it or it does not. Ask which figure the brand publishes for your caliber, then measure against it. That converts an opinion into a number.

Does hand-winding a new automatic damage it?

No. The mainspring in an automatic ends in a slipping bridle that lets go once the spring is full, so extra turns simply slip. Winding a new automatic to full before the seven-day test removes the single most common cause of a new watch losing time, and there is no wear argument against doing it daily.

My watch loses time only on the days I do not wear it. Is that a fault?

No, that is two normal effects stacking. An automatic left off the wrist runs down, and rate gets worse as the mainspring weakens. Resting in one fixed position for a day also removes the mixing of positions that a worn watch gets. Compare like with like: judge the rate over seven days of consistent daily wear, not over a week that included a stopped weekend.

Sources and further reading

  • Contrôle Officiel Suisse des Chronomètres (COSC), published certification criteria, for the minus 4 to plus 6 seconds a day mean daily rate and the seven-position, three-temperature test regime.
  • ISO 3159, Timekeeping instruments: wrist-chronometers with spring balance oscillator, for the rate measurement method the COSC criteria are built on.
  • Seiko published caliber specifications for the 4R and 6R families, for the plus 45 to minus 35 and plus 25 to minus 15 second daily bands and the conditions they are measured in.
  • Seiko Instruments technical data sheet for caliber NH35A, for the stated rate of minus 20 to plus 40 seconds a day and the 41 hour power reserve.
  • Rolex, Tudor and Omega published accuracy specifications for cased watches, including METAS Master Chronometer certification at 0 to plus 5 seconds a day.
  • Oris published specification for Calibre 400, at minus 3 to plus 5 seconds a day.
  • Published manufacturer warranty terms, which commonly run two to five years and cover manufacturing defects rather than normal rate variation.

Last reviewed 4 September 2026. Spotted an error? Tell us and we will fix it in public.