NHS Pay Scale Pay calculators & pay bands

Corrections

Any error that could have changed a figure someone acted on is published here, with what was wrong, who it affected, how large the error was, how we found it and what stops it happening again. Errors found before a figure was published are recorded in the change log instead.

Why publish these at all? Because a pay calculator that has never admitted an error either has not been used much or is not telling you something. The useful question is not whether a tool has made mistakes — it is whether you would find out.

  • We Said the Time Needed on Each Pay Step Is Not Published. It Is, in Every Nation, and We Told Readers to Distrust Sources That Gave It Correctly.

    misleading-statementfound 2026-09-17

    What was wrong: The increment dates page said the published pay circulars do not state the service required to move between pay steps, and that any site giving a table of years was 'telling you something its sources do not say'. The pay scale pages for every year repeated that the scales do not state how long each step takes, and the calculator's step assumption said the same. In fact NHS Employers prints 'years until eligible for pay progression' beside the England pay scales, as do the Welsh pay letter and the Northern Ireland pay circular, and Scotland publishes a year-by-year progression timetable in Table 1a of its section of the handbook.

    Who it affected: No pay, tax, pension or arrears figure was affected: the calculator asks for your step, not your start date. What was wrong was advice. Anyone trying to work out when their next step was due was told the answer was not published, and was steered away from accurate information. The statement was on the increment dates page and every pay scale year page until this correction.

    How it came to light: Reviewing why the increment dates page ranked for searches such as 'incremental date meaning' without being clicked. Reading the current NHS Employers pay scales page and Annex 2 of the NHS Terms and Conditions of Service Handbook to answer those searches showed the periods printed in the table itself.

    Fix: The periods for all four nations are now stored as sourced data and shown on the increment dates page in a table, with worked examples and a calculator that gives the date each step is due. The contractual rules on what resets or preserves the pay step date, and the standards for progression, are set out with handbook paragraph references. The sentence on the pay scale year pages and the calculator assumption were rewritten to give the periods.

    What stops it recurring: The periods live in one data file with a source for each nation. The table and the calculator both read it, so they cannot disagree, and the build stops if England, Wales and Northern Ireland diverge without the page being changed to show it. The broader lesson: a claim that something is not published needs the same check as a claim about what it says.

  • We Described a Formatting Typo on the NHS Employers Pay Scales Page as a Figure That Did Not Reconcile, and Named Them for an Error They Had Not Made.

    source-misattributedfound 2026-08-18

    What was wrong: NHS Employers print the England 2026-27 band 8d top hourly rate as £55,65, with a comma where the decimal point should be. The value is correct; only the punctuation is wrong. An earlier revision of our England 2026-27 file read that as £55.80, by dividing the annual £108,814 by 52 weeks × 37.5 hours. The page in fact uses about 1955.4 hours a year, which band 8c (£91,609 to £46.85) and band 9 (£129,783 to £66.37) both confirm. Because 55.80 does not reconcile with the annual figure, the build recorded it as a source discrepancy, and the sources page published the claim that NHS Employers had printed a rate implying a 1950.1-hour year.

    Who it affected: No calculated figure was wrong. The site already used the annual salary for every calculation and showed the derived £55.65, so take-home, pension and arrears results were unaffected in every band, nation and year. What was wrong was a published statement about a third party: anyone reading the sources page saw NHS Employers credited with an error of our own making.

    How it came to light: While preparing an email to report the comma to NHS Employers. Checking the arithmetic before sending showed the correction we had applied was itself wrong, and that the figure they print is right.

    Fix: The stored hourly is now 55.65, matching the source once its comma is read as a decimal point. The hourly-mismatch entry disappears with it, so the sources page now correctly reports that no published figure fails to reconcile. The comma remains an error on the source page and is still being reported to NHS Employers, as a formatting fix rather than a value change.

    What stops it recurring: A golden test now locks band 8d top at 55.65 so the ÷52 reading cannot return, and asserts that this case reports no mismatch. The wider lesson is recorded in the source file itself: where a source prints an unparseable figure, the divisor must be derived from neighbouring rows on the same page rather than assumed, and a transcription that departs from the source has to say which convention it used and why.

  • England Pension Contribution Tiers for 2023-24 Were the Previous Year’s, so Ten of Eleven Thresholds Were Wrong for a Whole Year.

    affected-figuresfound 2026-08-16

    What was wrong: The 2023-24 England and Wales tier table held the thresholds that applied in 2022-23. The uprating that took effect on 1 April 2023 had never been applied. Only tier 1 was correct, because its ceiling was frozen at £13,246 across both years. Every other boundary was too low, so a given salary fell into a higher tier than it belonged in.

    Who it affected: Anyone using a 2023-24 England or Wales figure whose pensionable pay sat between the old and new thresholds. The bands were wide, so this was not a boundary case: someone on £17,000 was charged 6.1% instead of 5.7%, about £68 a year; someone on £24,500 was charged 7.7% instead of 6.8%, about £220 a year; someone on £74,000 was charged 13.5% instead of 12.5%, about £740 a year. Take-home figures for that year were understated by the same amounts.

    How it came to light: A reader-supplied build brief contained a tier table that disagreed with ours for 2026-27. Checking that against NHS Employers confirmed our 2026-27 figures were wrong, which prompted a review of every year. The 2023-24 error surfaced when the published PDF for that year was read directly.

    Fix: All eleven tiers re-transcribed from the NHS Employers 2023-24 member contribution rates PDF. The 2022-23 table was separately confirmed against the DHSC consultation response and was already correct.

    What stops it recurring: A golden test now checks that thresholds actually move between consecutive years and that every threshold in an uprating moves by the same percentage. It needs no external source — a year that silently repeats the previous year’s figures now fails the build. Verified by reintroducing the error and confirming the test catches it.

  • Three England Pension Tier Boundaries for 2026-27 Were a Pound Too Low.

    affected-figuresfound 2026-08-16

    What was wrong: The 6.5%, 9.8% and 10.7% bands were recorded as ending at £28,853, £52,777 and £67,667. NHS Employers publishes them as ending at £28,854, £52,778 and £67,668. Our own Northern Ireland file, which shares England’s thresholds for this year, had them right, so the two files disagreed with each other.

    Who it affected: Anyone whose pensionable pay landed exactly on one of the three boundaries. Narrow, but costly where it hit: £28,854 was charged 8.3% instead of 6.5%, about £519 a year; £52,778 was charged 10.7% instead of 9.8%, about £475; £67,668 was charged 12.5% instead of 10.7%, about £1,218.

    How it came to light: A tier table supplied in a reader’s build brief disagreed with ours. Fetching the NHS Employers page settled it in the brief’s favour.

    Fix: Boundaries corrected to the published values. The rates themselves were always right.

    What stops it recurring: Every tier boundary for all three administrators is now tested from both sides — at the ceiling it must be this tier, a pound over it must be the next — with the published bands written into the test rather than read from the data. A test that reads the thing it is testing would have passed throughout.

  • Northern Ireland Pension Tiers for 2025-26 Followed HSC’s Webpage, Which Had Not Caught Up with an Amendment to Its Own Regulations.

    affected-figuresfound 2026-08-16

    What was wrong: The 2025-26 thresholds were taken from the HSC Pension Service compiled rates page and recorded as £27,288, £33,247, £49,913 and £63,994. Those figures were superseded by S.R. 2026/16, made on 2 March 2026 with retrospective effect to 1 April 2025, which sets them at £27,797, £33,868, £50,845 and £65,190 — matching England’s thresholds for that year, with Northern Ireland’s own higher rates. The HSC page still shows the superseded figures.

    Who it affected: Northern Ireland results for 2025-26. Someone on £27,500 was charged 8.5% instead of 6.7%, about £495 a year; £33,500 was charged 10.0% instead of 8.5%, about £502; £64,500 was charged 12.7% instead of 10.9%, about £1,161.

    How it came to light: An internal consistency check flagged it before any source did. Northern Ireland’s stored thresholds implied a 5.74% uprating into the verified 2026-27 figures where England implied exactly 3.80%. A reader then supplied the statutory text that explained the discrepancy.

    Fix: Thresholds taken from the consolidated statutory text, regulation 30(3A) of S.R. 2015/120 as amended. The rates were already correct. The conflict with the administrator’s own page is recorded in the data file rather than quietly resolved.

    What stops it recurring: The uprating-consistency check that caught this is now a permanent golden test. Sourcing practice has also changed: where consolidated statutory text exists on legislation.gov.uk, it is preferred over an administrator’s summary page, because summary pages lag their own legislation.

  • Northern Ireland 2023-24 Tiers Overlapped by £10, Reproducing an Error in the Published Source.

    affected-figuresfound 2026-08-16

    What was wrong: Tier 4 was recorded as ending at £25,156 while tier 5 began at £25,147, so the two overlapped and anyone paid in that range fell into both. This was copied faithfully from the HSC Pension Service published table, which still shows the same overlap. Tier 4 should end at £25,146: that is what HSC’s own tier 5 start implies, and what NHS Employers publishes for the identical England thresholds that year.

    Who it affected: Northern Ireland results for 2023-24 with pensionable pay between £25,147 and £25,156. Those salaries were charged 6.8% when the correct reading is 7.7%. A narrow band, and the error was in the source rather than in transcription, but the figure shown was still wrong.

    How it came to light: Comparing the Northern Ireland table against the England PDF for the same year, which shares its thresholds. Ten of eleven matched exactly and one did not.

    Fix: Tier 4 read as ending at £25,146 and tier 5 as beginning at £25,147. The discrepancy in the source is recorded in the data file, not silently corrected.

    What stops it recurring: A golden test now asserts that tiers tile the whole pay range with no gaps and no overlaps, for every administrator and every year. It needs no external source and would have caught this on the day the figures were entered.

  • Pension Contribution Tiers Were Wrong for Northern Ireland in All Five Years, and for Scotland in Two.

    affected-figuresfound 2026-08-08

    What was wrong: Northern Ireland: 2022-23 and 2023-24 were stored as a six-tier structure with a 12.7% top rate. Both years actually ran the eleven-tier structure with a 13.5% top rate. The 2024-25 entry held what are in fact the 2025-26 figures, a year out. 2025-26 and 2026-27 held thresholds that had been uprated a further time beyond the published tables. Scotland: the 2025-26 thresholds had been uprated by 4.25%, the Agenda for Change award before an inflation guarantee lifted it to 4.4%, and were never revised. 2026-27 inherited the error from that base. Scotland 2026-27 was also flagged as awaiting a circular when SPPA circular 2026/03 had been published on 10 March 2026.

    Who it affected: Every Northern Ireland result in every year, and every Scottish result for 2025-26 and 2026-27. The size of the error depended on where a salary sat. Most were unaffected in cash terms, because a wrong threshold and a right one give the same answer unless a salary sits between them. Where it did bite it was substantial: a Band 7 entry in Northern Ireland for 2023-24 was charged 7.7% instead of 9.8%, understating the pension deduction by about 919 pounds a year and overstating take-home by about 735. England and Wales results were not affected at any point.

    How it came to light: A request to add the missing source URLs for the two schemes. Reading the publishers' own pages to find those links meant reading the contribution tables printed on them, which did not match our stored figures.

    Fix: Both files were replaced with the tables published by the Scottish Public Pensions Agency and the HSC Pension Service, read on 8 August 2026. Every year now records the document it came from, its URL and the date it was retrieved. Scotland 2022-23, 2023-24 and 2024-25 figures are unchanged: their circulars are linked and resolve, but those PDFs are not text-extractable, so the tables could not be independently re-read and are marked unverified rather than silently endorsed.

    What stops it recurring: The test corpus had 22 cases covering 157 figures and did not catch any of this, because every case happened to sit comfortably inside a tier rather than on the edge of one. Seven cases were added that stand on the boundaries: two a single pound either side of a Scottish tier ceiling, and four pinning the Northern Ireland structure and rates in each year. Verified by running them against the old data, where they fail. Separately, a named source without a link is now a hard test failure rather than a declared exception, which is what surfaced this in the first place.

  • High Cost Area Supplement Floors and Ceilings Were Wrong for Four of the Five Pay Years.

    affected-figuresfound 2026-08-08

    What was wrong: The stored floor and ceiling values for inner London, outer London and the London fringe did not match the HCAS tables published alongside each year's pay scales, for 2022-23, 2023-24, 2024-25 and 2025-26. The 2026-27 values were correct.

    Who it affected: Any England result for those four years with a London supplement selected, where the supplement fell on its floor or its ceiling. The error ranged from £150 to £460 a year depending on zone and year. Results with no supplement, results for 2026-27, and results where the supplement fell between the floor and the ceiling were all unaffected.

    How it came to light: A build-time cross-check. Each year's England pay file also ships the whole-time HCAS table, so the floor and ceiling can be recovered by subtraction from about a hundred independent cells. Those recovered values disagreed with the stored ones.

    Fix: The stored values were regenerated by subtraction from the published tables.

    What stops it recurring: The build now derives the floors and ceilings from the published tables on every run, reports any disagreement with the rules file, and stops the build outright if the rule cannot reproduce every published HCAS figure. The same figures cannot drift apart again without the build saying so.

Report an Error

If a figure disagrees with your payslip or with a published source, it is worth reporting even if you are not sure — a report that turns out to be a misunderstanding still tells us the explanation is not clear enough.

Useful details: your band and step, nation, pay year, contracted hours, London zone, whether you are in the pension scheme, and what you expected to see.

Report it to contact@nhspayscale.co.uk. Every report is read. If it turns out a figure is wrong, the correction is published here with what was wrong, who it affected and how large the error was — including when the mistake was ours, which most of the entries below are.