UK employers give a median of 24 days' annual leave, excluding bank holidays. Half of all organisations sit between 22 and 26 days. A tenth stop at 20 days — and no organisation in the panel goes lower — while the top tenth gives 30 or more, stretching to 35 (lumi panel, n=261, August 2026).
Key findings
- The median entitlement is 24 days excluding bank holidays; the interquartile range runs 22–26 days (n=261).
- 44% of organisations add extra annual leave for length of service; 56% do not (n=257).
- Half the panel (50%) lets employees buy extra leave, but only 14% can sell any back — and unlimited leave is rare: 94% do not offer it to anyone (n=258; n=267).
- 76% allow some carry-over of unused days; at 24% of organisations unused leave is forfeited (n=252).
How many days of annual leave is typical in 2026?
The median entitlement across the panel is 24 days, excluding bank holidays. Entitlements are recorded on that basis — days of leave beyond the 8 UK bank holidays — so organisations that package bank holidays differently can be compared like for like (n=261).
The corridor is narrow but real. The bottom tenth of the panel sits at 20 days — none gives fewer. At the other end, the top tenth gives 30 days or more, and the most generous organisation reports 35.
Two things stand out. The distribution leans upwards: the bottom tenth sits four days below the median, while the top tenth sits six above it — variation among UK employers is mostly a question of how far above 24 they go. And the floor is hard: no organisation reports fewer than 20 days. Statutory minimum leave leaves little room below that in any case.
Does annual leave grow with length of service?
Often, yes. 44% of organisations provide additional annual leave based on length of service; 56% do not (n=257). That split matters for benchmarking: two organisations advertising the same day-one entitlement can diverge within a few years, and a middle-of-the-panel offer at hire may sit in the upper half for long-serving staff.
The panel records whether entitlement rises with service, not by how many days. Treat the uplift itself — its size, and the tenure points at which it lands — as a separate design decision rather than something the headline figure settles.
Can employees buy or sell annual leave?
Half the panel offers a route to more days. 50% of organisations let employees buy extra leave — 38% through a buy-only scheme and 12% through a scheme that allows both buying and selling. Selling is much rarer: 14% allow it in any form, and only 2% run a sell-only scheme (n=258).
| Can employees buy or sell leave? | Share of panel | Organisations |
|---|---|---|
| Buy only | 38% | 98 |
| Both buy and sell | 12% | 31 |
| Sell only | 2% | 5 |
| No scheme | 48% | 124 |
Two things stand out. Buying is mainstream — a low-cost lever, since the employee funds the extra days — and half the panel has adopted it. Selling is an afterthought: where it exists it almost always rides alongside a buy option, and sell-only schemes amount to five organisations in the whole panel.
How common is unlimited annual leave?
Rare — and it is worth saying so plainly. 94% of the panel does not offer unlimited leave to anyone. A further 4% offer it for some roles only, and 2% — five organisations — extend it to all employees (n=267). Taken together, 6% of UK organisations offer uncapped leave in any form. On these figures it is a niche, not an emerging norm.
Can unused leave be carried over?
Mostly yes, but on conditions. 47% of organisations allow limited carry-over in specific circumstances, and 29% run standard carry-over. At the remaining 24%, unused leave is forfeited (n=252). Put the two "yes" answers together and 76% of the panel lets unused days travel in some form — but fewer than a third make it a standing right rather than an exception.
How do employers treat bank holidays?
There is no standard answer. 40% of organisations provide bank holidays in addition to the annual leave allowance, 39% count them within it, and 22% run a mix depending on role or contract (n=215). This is why every figure on this page is stated excluding bank holidays: with the 8 UK bank holidays in play, the same headline number can describe offers eight days apart. A benchmark only works once every offer is restated on the same basis.
What should a reward team do with this?
- Compare on one basis. Restate every offer as days excluding bank holidays before benchmarking. With the panel split 40% on top, 39% within and 22% mixed on packaging, a raw days figure is ambiguous until you know what it includes.
- Treat 22–26 as the corridor. The middle half of the panel spans four days. The real decision is where in that band you sit at day one — and whether service days, as at 44% of the panel, do the differentiating after that.
- Use flexibility before headcount-wide days. A buy scheme — already in place at half the panel — flexes individual entitlement without moving the baseline; unlimited leave, at 6% of the panel, is not the benchmark to chase.
About these figures
Figures are medians and quartiles computed across the lumi benchmarking panel — a reference panel of UK organisation profiles modelled from graded published UK survey sources and calibrated metric by metric against lumi's anchor register, growing as member organisations contribute their own data. Each organisation reports one headline figure for its main population; comparisons are medians and percentiles, never averages, and any figure resting on fewer than 5 organisations is suppressed.
Sample sizes are stated per figure and were computed in August 2026. Read the full methodology.