GMT vs UTC: Why They Show the Same Time but Are Not the Same Thing

Put a clock labelled GMT next to a clock labelled UTCand they will agree on the hour, the minute, and — to within a fraction of a second — the second. That is why almost everyone treats the two as interchangeable, and why almost nothing goes wrong when they do. But they are not the same kind of thing. GMT is a time zone: a label for civil time on a particular strip of the map, set by the governments that live on it. UTC is a time standard: the atomic reference from which every zone offset in the world, GMT included, is defined. Swapping the words is harmless in a text message and expensive in a flight plan, a trading system, or a database schema. This article covers what each one actually is, who computes UTC and how, why leap seconds exist and what the world's metrology bodies have decided to do about them, the British Summer Time trap that makes "GMT" wrong for more than half the year in the country that invented it, what "Zulu" means, and how software should store zero-offset time. If you only need the arithmetic, the timezone converter will do it for you.

The answer in one sentence

GMT (Greenwich Mean Time) is a time zone at UTC+0. UTC (Coordinated Universal Time) is the standard that defines what "UTC+0" means. On a clock face they read identically. In a specification they belong in different rows. Everything else below follows from that single distinction.

Another way to see it: asking "what time is it in GMT?" has a wall-clock answer, because GMT is somewhere people live and work. Asking "what time is it in UTC?" is really asking for the reference itself. UTC is not a place. It belongs to no country, no parliament can move it, and — this is the part people forget most often — it never observes daylight saving time. It does not shift, ever.

What UTC actually is

UTC is not read off a single master clock in a single building. It is a computed time scale. Roughly 450 atomic clocks held by around 85 national metrology institutes and observatories send their data to the Bureau International des Poids et Mesures (BIPM) at Sèvres, outside Paris. The BIPM combines those readings into a free-running weighted average, then steers that average using the world's primary frequency standards — caesium fountains that realise the SI second as accurately as it can currently be realised. The result is International Atomic Time (TAI), a continuous count of SI seconds with no breaks in it at all.

UTC is then derived from TAI by subtracting a whole number of seconds:

• UTC = TAI − 37 s, and has been since 1 January 2017.
• That integer changes only when a leap secondis inserted, which keeps UTC in step with the Earth's actual rotation rather than letting it run off into pure atomic time.

Two consequences follow that surprise people. First, UTC is published in deferred time: the BIPM issues Circular T monthly, giving the difference [UTC − UTC(k)] at five-day intervals for each contributing institute. Nobody can hand you UTC in real time. What an NTP server, a GPS receiver or a national time signal actually gives you is a local physical realisation — UTC(NPL), UTC(NIST), UTC(USNO) and so on — each of which tracks UTC to within nanoseconds but is not literally UTC. Second, because UTC is an atomic scale with corrections applied, it is not the same thing as the Earth's rotation angle; that quantity has its own name, UT1, and it is what leap seconds exist to chase.

What GMT actually is

Greenwich Mean Time began as an astronomical quantity: mean solar time on the meridian of the Royal Observatory, Greenwich. The word meanis doing real work there. Actual solar days vary in length across the year — the discrepancy between a sundial and a clock, the equation of time, reaches roughly a quarter of an hour — so the standard was defined against a fictitious sun moving at a uniform rate. The 1884 International Meridian Conference in Washington adopted the Greenwich meridian as the world's prime meridian, and Greenwich time became the reference from which other countries counted their offsets.

The term then went through a rename that still causes trouble in old documents. Astronomers had counted the GMT day from noon, not midnight, so that a single night's observing run fell on one date. From 1 January 1925 the astronomical day was shifted to begin at midnight, matching civil usage, and in 1928 the International Astronomical Union recommended the name Universal Time for the Greenwich-meridian scale in almanacs, precisely to escape the ambiguity. The scientific successor to GMT is therefore UT1, not UTC.

What survives under the name "GMT" today is a civil time zone. It is the winter clock setting in the United Kingdom and Ireland, the year-round setting in Iceland, and the year-round setting across a stretch of West Africa including Ghana, Côte d'Ivoire, Senegal, Mali, Burkina Faso, Guinea, Sierra Leone, Liberia, the Gambia, Mauritania and Togo. Portugal and the Canary Islands sit on the same offsets but label them WET and WEST rather than GMT and BST. Being a civil zone, GMT answers to legislatures: any of those countries could move its clocks tomorrow and GMT's membership list would change. UTC would not notice.

Leap seconds and the decision to stop them

The Earth is not a good clock. Tidal friction slows its rotation over the long term, and shorter-term wobbles from the atmosphere, the oceans and the fluid core add irregular noise on top. Atomic time does not care about any of that, so UT1 and TAI drift apart. Leap seconds are the patch.

Recommendation ITU-R TF.460 fixes the tolerance: UTC must be kept within 0.9 secondsof UT1. The International Earth Rotation and Reference Systems Service (IERS) monitors the gap and announces insertions about six months ahead in its Bulletin C. A leap second is added as an extra 61st second at the end of a UTC month, preferentially June or December, so the sequence runs 23:59:59 → 23:59:60→ 00:00:00.

The record, stated precisely, is this:

• UTC took its present stepped form on 1 January 1972, when TAI − UTC was set to exactly 10 seconds.
• 27 leap seconds have been inserted since. Every one of them has been positive— a second added, never removed. 10 + 27 = the 37-second offset in force today.
• The most recent was inserted at the end of 31 December 2016. There has not been one since, and the IERS has confirmed there will be none at the end of 2026 either. That is already the longest run without a leap second since the system began.

In November 2022 the 27th General Conference on Weights and Measures adopted Resolution 4, "On the use and future development of UTC". This is the single most misreported fact in the topic, so it is worth stating exactly. The resolution does not abolish UTC, and it does not declare leap seconds illegal from a fixed date. What it decides is that the maximum permitted value of (UT1 − UTC) will be increased in, or before, 2035. It asks the CIPM to consult the ITU and other affected bodies, propose a new tolerance large enough to keep UTC continuous for at least a century, prepare an implementation plan, and bring a resolution to the 28th CGPM in 2026. Raise the tolerance far enough and leap seconds simply stop being needed — but the replacement value and the exact date are decisions still being taken, not settled history.

Part of the urgency is that the Earth has recently been rotating slightly faster, which raises the prospect of a negativeleap second — deleting 23:59:59 rather than inserting 23:59:60. That has never been done. Very little deployed software has ever been tested against it, which is a large part of why the timing community would rather retire the mechanism than exercise it in a direction nobody has rehearsed.

The GMT-versus-BST trap

Here is where the distinction stops being academic. The United Kingdom does not stay on GMT. It observes British Summer Time (BST), UTC+1, from the last Sunday in March to the last Sunday in October, with both transitions occurring at 01:00 UTC. For roughly seven months of the year, London is not on GMT.

British usage has not kept up. "GMT" is routinely used in Britain to mean "UK time, whatever it currently is" — in emails, on broadcast schedules, in meeting invitations, on websites that were built in January and never revisited. So when a London colleague writes "let's meet at 10:00 GMT" in July, there are two readings:

• Literal: 10:00 UTC+0, which is 11:00 on their own wall clock. Almost never what they meant.
• Intended: 10:00 on the London wall clock, which is BST, which is 09:00 UTC.

That is a clean one-hour error, and it is the single most common mistake in this whole subject. The safe habit is to treat "GMT" from a British or Irish sender between late March and late October as an unresolved question and ask, or to convert from a named city rather than an abbreviation.

Ireland adds a wrinkle worth knowing about if you write software. Under the Standard Time (Amendment) Act 1971, Irish Standard Time is UTC+1 and is the standard, with GMT observed in winter as the departure from it. The tz database models this faithfully: Europe/Dublin carries a negativedaylight saving offset, so its "daylight" setting is one hour behind its standard, and the abbreviation IST is properly read as Irish Standard Time. Dublin and London show the same clock all year; their bookkeeping is inverted.

Zulu time

Aviation, the military and a good deal of computing avoid the whole argument by using a third label for the same instant. In the military time-zone lettering system, each hourly offset gets a letter, and the zero-offset zone gets Z. Spoken using the NATO phonetic alphabet, Z is Zulu — so Zulu time is UTC, exactly and without qualification. Twelve hundred Zulu is 12:00 UTC. (The same system uses J, "Juliett", for the observer's own local time.)

The convention carries into text formats. ISO 8601 — and RFC 3339, the profile of it used across the web — write a trailing Z to mean an offset of exactly +00:00, as in 2026-07-29T12:00:00Z. Flight plans, METAR and TAF weather reports, NOTAMs, and military operation orders are all timed in Zulu for the obvious reason: a flight crossing six time zones needs one clock that nobody has to reason about, and an aircraft has no interest in whether the ground below it has just sprung forward.

Quick reference: 12:00 UTC around the world

The table below shows what noon UTC looks like elsewhere, in the northern winter and northern summer. Note that UTC is identical in both columns — that is the whole point of it. Note also that the European and North American daylight saving schedules do not switch on the same dates, so for a couple of weeks each spring and autumn the transatlantic rows are one hour off what the table implies.

How software should handle it

The rule that follows from everything above: store the instant in UTC, store the location as an IANA zone name, and never store the string "GMT" as a zone.

The tz database makes the difference explicit. Etc/UTC is a fixed zero-offset zone with no rules attached; it will never move, which is exactly what you want for a machine timestamp. Europe/London is a civil zone that carries the full GB-Eire and EU rule history and renders as GMT or BST depending on the date; that is what you want for a person in Manchester. A bare GMT resolves to Etc/GMT, a fixed UTC+0 zone with no summer time at all — correct for a protocol, silently wrong for eight months a year for a human user. The database also ships Etc/Zulu and Etc/Universal as links to Etc/UTC.

One trap in that family deserves a warning of its own. The Etc/GMT±N zones follow the POSIX sign convention, which counts positive westof Greenwich — the opposite of the way offsets are written everywhere else. So Etc/GMT+5 is UTC−5, not UTC+5, and Etc/GMT-5 is UTC+5. If you find that pair in a config file, assume it is wrong until you have checked.

Leap seconds bring a separate complication: POSIX time, and therefore Unix timestamps, has no way to represent 23:59:60. A day is defined as exactly 86,400 seconds. Systems cope by repeating or freezing a second, or by smearing— large cloud providers spread the extra second across many hours as a tiny frequency offset, so no clock ever shows an impossible value. If your code does precise interval arithmetic across a leap second boundary, you need to know which strategy your platform uses.

Worked examples

Example 1 — a London meeting, twice a year. A standup at 09:00 on the London wall clock. On 14 January, London is on GMT, so the instant is 09:00 UTC. On 8 July, London is on BST, so the same wall-clock time is 08:00 UTC. Nothing in the invitation changed; the UTC instant moved by an hour. Anyone joining from a non-European zone sees the meeting shift. Compare it against the GMT to UTC conversion if you want the offsets side by side.

Example 2 — reading a Zulu timestamp. A weather report stamped 291200Zmeans the 29th of the current month at 12:00 UTC. For a crew in New York in July that is 08:00 EDT; in January it is 07:00 EST. The report itself never changes, because Zulu never changes — see the UTC to EST conversion for the winter figure.

Example 3 — GPS time is not UTC.This one catches engineers. GPS system time runs continuously with no leap seconds, and is offset from atomic time by a constant: TAI − GPS = 19 s. Since TAI − UTC = 37 s, it follows that GPS − UTC = 18 s. A receiver that reports raw GPS time is eighteen seconds ahead of the wall clock, and that gap has grown every time a leap second was inserted. Most receivers apply the correction for you using the offset broadcast in the navigation message; some do not.

Example 4 — converting a US business hour to Greenwich.A 09:00 Pacific call in January is 17:00 UTC (UTC−8), which is 17:00 GMT in London — a workable if late 5 p.m. In July the same call is 09:00 PDT, so 16:00 UTC, which London reads as 17:00 BST. Same UK wall-clock time, a different UTC instant, because both ends moved. The UTC to PST conversion spells out the winter side of that.

Common mistakes

• Writing "GMT" in a technical specification. If you mean the standard, write UTC. GMT is a civil zone and its membership and rules are subject to national law.
• Reading a British "GMT" literally in summer. From late March to late October, a UK sender who writes GMT almost always means BST, which is UTC+1. One hour, every time.
• Expecting UTC to change for daylight saving. It never has and never will. If your clock labelled UTC moved in spring, it was not UTC.
• Trusting the sign in Etc/GMT+N. The POSIX convention inverts it. Etc/GMT+5is UTC−5.
• Treating GMT and UT1 as the same quantity.UT1 is the Earth's rotation angle expressed as a time scale — the scientific descendant of astronomical GMT. Civil GMT is a zone label pinned to UTC+0. They are usually within a fraction of a second of each other, which is precisely why the difference is easy to miss and awkward when it matters.
• Believing leap seconds are already gone.The 2022 CGPM resolution commits to raising the UT1 − UTC tolerance in or before 2035 and to bringing a concrete proposal to the 2026 CGPM. Until that lands, the mechanism is still in force — the IERS still publishes Bulletin C, and it still could announce one.
• Storing a fixed offset instead of a zone. +00:00 in a user record is a snapshot, not a rule. Store Europe/London and let the library decide whether today is GMT or BST.

Bottom line

GMT and UTC agree on the reading and disagree on the category. GMT is a time zone — historically mean solar time at the Royal Observatory, Greenwich, today the civil clock of the UK and Ireland in winter, Iceland and much of West Africa year-round. UTC is the standard those zones are measured against: computed by the BIPM from roughly 450 atomic clocks at about 85 institutes, derived from TAI by an integer offset that has stood at 37 seconds since January 2017, and held within 0.9 seconds of the Earth's rotation by leap seconds that the CGPM has resolved to render unnecessary in or before 2035. UTC does not observe daylight saving. GMT's users do. In practice: say UTC when you mean the standard, say Zulu if you are flying, ask what the sender meant when a Briton says GMT in July, and in code store an instant in UTC alongside an IANA zone name — never the three letters G, M, T.