What dates a week covers is the second thing anyone asks, and it is a question about the whole year, not one week. Weeks run Monday to Sunday; — of them this year.
Two things vary between conventions: which day starts a week, and which week counts as week 1. The broadcast convention starts on a Monday exactly like ISO does and still disagrees with it — so it is the week-1 rule that does the work, not the start day.
—
—
An ISO year has 53 weeks when 1 January falls on a Thursday, or on a Wednesday in a leap year — because week 1 is defined as the week containing the year's first Thursday, and those are the two ways to fit an extra one in. Everything else has 52.
ISO 8601 week 1 is the week containing the year’s first Thursday: the same as the week containing 4 January. Weeks run Monday to Sunday.
The year holding its Thursday. So 29–31 December can sit in week 1 of the next year and 1–3 January in week 52 or 53 of the last.
ISO writes 2026-W01. A week number stored without its ISO year loses information for six days a year.
When 1 January is a Thursday, or a Wednesday in a leap year: about every five or six years. 2026 is one.
It is the week containing the first Thursday, the same as the week containing 4 January and the first week with most of its days in the new year. If 1 January is a Friday, it sits in the previous year’s last week.
Six days a year belong to a different week-year than calendar year. That is why ISO writes 2026-W01. Systems that store just the number quietly corrupt those days.
WEEKNUM defaults to Sunday-start with week 1 containing 1 January; ISOWEEKNUM follows ISO. Two functions, one spreadsheet, different answers for the same date.
Weekly payroll gets a 53rd run and retail calendars gain a trading week. They come every five or six years, and finance systems built on “52 weeks” find out the hard way.
A local-time date crossing a daylight-saving boundary can land on the wrong side of midnight and shift into an adjacent week. Twice a year, in one timezone, for one hour: the hardest kind of bug to reproduce.
Some systems emit “week 0” for early-January days that ISO assigns to the previous year’s final week, a symptom of implementing the numbering without the year pairing.
Because it is probably not using ISO. Excel's WEEKNUM defaults to weeks starting
Sunday with week 1 being whichever week contains 1 January — a different rule that disagrees with
ISO for several weeks most years. ISOWEEKNUM is the one that matches this tool.
Those where 1 January is a Thursday, or a Wednesday in a leap year. They arrive roughly every five to six years and matter more than they sound: weekly payroll gets a 53rd pay run, and retailers running 52-week financial calendars have to absorb an extra trading week.
No. ISO 8601 is standard across Europe and much of business, but the US, Canada and parts of Asia commonly start weeks on Sunday and number from the week containing 1 January. Both are shown above, because agreeing which convention you mean is most of the work.
That a date is a date. Everything here is computed in UTC with no time of day, which is deliberate: week numbering done in local time can put the same instant in two different weeks either side of a time-zone boundary, and midnight arithmetic across a daylight-saving change is where most week-number bugs live.
Two conventions are covered — ISO 8601 and the US Sunday-start — and not the
Saturday-start weeks used in parts of the Middle East, nor the several spreadsheet variants
beyond Excel’s default. The ISO figures have been checked against an independent
reference implementation on every day from 1900 to 2100, all 73,414 of them, with no
disagreement in either the week number or the ISO year: weeknumber_model.py.
Week numbering itself has no meaning before the Gregorian calendar was adopted locally,
which happened at different dates in different countries.
Not an ISO concept. Some systems emit "week 0" for early-January days that ISO assigns to the previous year's final week — a symptom of implementing the numbering without the year pairing.
A week belongs to whichever year holds most of it — formally, the year containing that week’s Thursday. Since 52 weeks is only 364 days, the two calendars cannot stay in step, and the seam falls across New Year.
A year gets a 53rd week when 1 January is a Thursday, or a Wednesday in a leap year. It happens about every five or six years.
2026 is one of them. Payroll, rotas and reporting periods built on “52 weeks” are a week short this year, which is the sort of thing that is noticed in December rather than January.
| Date | Calendar year | ISO week | Falls in |
|---|---|---|---|
| Mon 30 Dec 2024 | 2024 | 2025-W01 | the year after |
| Tue 31 Dec 2024 | 2024 | 2025-W01 | the year after |
| Mon 29 Dec 2025 | 2025 | 2026-W01 | the year after |
| Tue 30 Dec 2025 | 2025 | 2026-W01 | the year after |
| Wed 31 Dec 2025 | 2025 | 2026-W01 | the year after |
| Fri 01 Jan 2027 | 2027 | 2026-W53 | the year before |
| Sat 02 Jan 2027 | 2027 | 2026-W53 | the year before |
| Sun 03 Jan 2027 | 2027 | 2026-W53 | the year before |
| Sat 01 Jan 2028 | 2028 | 2027-W52 | the year before |
| Sun 02 Jan 2028 | 2028 | 2027-W52 | the year before |
This is the trap. 1 January 2027 is in week 53 of 2026, and 30 December 2024 is already in week 1 of 2025. A spreadsheet that pairs an ISO week number with the calendar year of the date will silently file those rows in the wrong year.
Releases in which this page changed, newest last. Derived from the archived copy of every release, not from notes written afterwards — so it reflects what actually shipped. Site-wide passes are left out; they are in the full changelog.