Withdrawal Sequencing Around a Large One-Off Expense

Key takeaways

• Taking £120,000 from a defined contribution pension in one tax year, with no other income, leaves £23,432 of income tax on 2026 to 2027 English bands.
• Spreading the same £120,000 over six tax years leaves £2,916. The gap is £20,516, and the effective rate on the pot falls from 19.5% to 2.4%.
• On top of a £60,000 salary, a £90,000 taxable withdrawal costs £42,271 in one year against £36,000 spread over three, because of the 60% band above £100,000.
• A withdrawal that fills the basic rate band moves gains on shares from 18% to 24%. On a £17,000 chargeable gain that is £1,020 of extra tax.
• HMRC repaid £50,353,656.76 on 12,612 flexible pension overpayment claims between 1 April and 30 June 2026, an average of £3,993 a claim.

The same £120,000 costs £23,432 in one tax year and £2,916 across six

You have a large one-off bill coming. A new roof, a care home deposit, a wedding, a house deposit for a child. The money sits in a pension, and you want to know what taking it in one lump costs you compared with taking it in pieces. That question is withdrawal sequencing, and in the UK it has a number attached to it.

Here's the size of it. Take £120,000 from an uncrystallised defined contribution pot in the 2026 to 2027 tax year, with no other taxable income. Up to 25% comes out tax free, capped at £268,275 across all your pensions, so £30,000 of that withdrawal carries no tax. The other £90,000 is taxable income in that year. Set against a £12,570 personal allowance, a basic rate band running to £50,270 at 20%, and 40% above that, the bill is £23,432.

Now take £20,000 a year for six years instead. Each year, £5,000 is tax free and £15,000 is taxable. Each year the personal allowance absorbs £12,570 of it and £486 of tax falls due on the rest. Six years of that is £2,916.

Same pot. Same £120,000. A difference of £20,516, which is the whole of the answer to the question. The effective tax rate on the money withdrawn drops from 19.5% to 2.4%.

Withdrawal sequencing is a question about how many allowances and bands you get to use

There's nothing clever going on in that arithmetic. The personal allowance and the basic rate band are annual, they don't carry forward, and an unused one is gone at midnight on 5 April.

The chart below plots the same £120,000 spread across one, two, three, four and six tax years. Two years costs £12,972. Three years costs £10,458. Four years costs £7,944. The curve is steepest at the start, because the first thing spreading buys you is escaping the 40% band, and only after that does it start buying extra personal allowances.

That shape matters more than the headline number. Most of the saving on this example, £10,460 of the £20,516, arrives by moving from one tax year to two. Someone who can't wait six years but can wait one has already captured half of what's available.

The same logic runs in reverse for anyone whose income is already high. If your other income has used the personal allowance and filled the basic rate band before the withdrawal starts, spreading buys nothing at the bottom of the scale. It only helps where the withdrawal would otherwise cross a threshold above you.

Above £100,000 the personal allowance taper creates a 60% band that lumpy withdrawals fall into

The single most expensive feature of the UK income tax system for a one-off withdrawal isn't the 45% additional rate. It's the taper. GOV.UK puts it plainly: "Your personal allowance goes down by £1 for every £2 that your adjusted net income is above £100,000." The allowance reaches zero at £125,140.

Work through what that does to a pound of income in that stretch. The pound itself is taxed at 40%. It also costs you 50p of personal allowance, which was previously untaxed and is now taxed at 40%, adding 20p. So the pound costs 60p. The marginal rate between £100,000 and £125,140 is 60%, It appears in no rate table, because it's an emergent property of two rules rather than a rate anybody legislated.

Put a £90,000 taxable pension withdrawal on top of a £60,000 salary and the whole of that stretch gets used. Total income is £150,000. The withdrawal is taxed at 40% up to £100,000, at an effective 60% from there to £125,140, and at 45% above. The tax attributable to the withdrawal is £42,271, or 47.0% of it.

Split the same £90,000 into three annual slices of £30,000 on top of the same salary and each year's income is £90,000, comfortably below the taper. The withdrawal costs £12,000 a year, £36,000 in total, a flat 40.0%. The £6,271 difference is bought entirely by staying under £100,000.

A large pension lump sum also raises the tax rate on gains you realise elsewhere

This is the interaction people miss, because the two taxes look unrelated. Capital gains tax on shares is not a flat rate. HMRC's rule adds your gains to your taxable income: within the basic rate band you pay 18% on gains made from 6 April 2026, and above it you pay 24%.

So a pension withdrawal that consumes your basic rate band pushes every gain realised in the same tax year from 18% to 24%. Suppose you also sell £20,000 of gains from a taxable account to help fund the expense. The £3,000 annual exempt amount takes the chargeable gain to £17,000. At 18% that's £3,060. At 24% it's £4,080. Selling in a year when the pension withdrawal has filled the band costs £1,020 more than selling in a year when it hasn't.

That number is small next to the income tax figures, and it's included precisely because it's small. Withdrawal sequencing changes the rate on the other pots too, not just on the pension, and the effect compounds quietly. The UK investment tax rates that apply to each pot are what make the order of withdrawal worth thinking about at all.

The ranking in that table is the reason asset location and decumulation are the same problem viewed from opposite ends. Where a pound sits determines what it costs to get it out, which is the same arithmetic that drives the ISA vs pension comparison on the way in.

Whatever the sequencing, the first payment usually comes out short

There's a mechanical wrinkle that catches people who've done the arithmetic correctly and still get less than they expected. When a scheme administrator has no PAYE code for you, HMRC's manual says they deduct "tax from the payment using the emergency tax code on a week 1 / month 1 basis". The emergency tax code for 2026 to 2027 is 1257L M1.

Month 1 means the payment is taxed as though it were the first of twelve identical monthly payments. A single £90,000 taxable withdrawal is treated as the opening instalment of a £1,080,000 annual income. The over-deduction is refundable, either in year through form P55, P53Z or P50Z, or automatically after the tax year ends.

The scale of this isn't hypothetical. HMRC's July 2026 pension schemes newsletter reports that between 1 April 2026 and 30 June 2026 it processed 10,200 P55 forms, 2,001 P53Z forms and 411 P50Z forms, repaying £50,353,656.76. That's 12,612 claims in a single quarter, averaging £3,993 each, and it only counts the people who filled in a form.

None of that changes the eventual tax. It changes when you have the money, which is the entire point if the withdrawal is funding a bill with a date on it.

The objection to spreading: the expense doesn't wait, and the first withdrawal costs you anyway

The strongest objection to everything above is that the arithmetic assumes you have the option. You usually don't. A roof, a care fee or a completion date arrives on its own schedule, and a tax saving spread over six years is worth nothing to a bill due in March.

Bridging the gap has a price, and it's the price the tax saving has to beat. On the six year example, £20,516 spread across five years of waiting is roughly £4,100 a year of borrowing cost that spreading can absorb before it stops paying. Whether that's a lot depends on the rate you'd be charged and on nothing else.

The second objection is sharper, because it applies to the small withdrawal as much as the large one. HMRC's pensions tax manual records that "The money purchase annual allowance for tax year 2023 onwards is £10,000", and it's triggered by flexibly accessing a money purchase pension. Once triggered it applies for that year and every year after. Someone still working and still contributing goes from a £60,000 annual allowance to £10,000, and takes that hit on the first £20,000 slice exactly as they would on a single £120,000 lump. Spreading doesn't soften it, because the trigger fires once and then applies to every year that follows.

The third objection is that money left in the pension stays invested. Six years of spreading is six years of market exposure on the balance, which cuts both ways, and it's the same asymmetry that makes sequence risk matter at the start of retirement. A market fall in year two doesn't care that you saved £20,516 in tax.

What this arithmetic cannot tell you

Start with the biggest limitation. Every figure above applies 2026 to 2027 English and Northern Irish rates to all six years, and there's no basis for assuming they hold. Six years of spreading is a bet on six Budgets. The personal allowance, the basic rate band and the taper threshold are all set annually.

The rates aren't even uniform across the UK today. Scottish taxpayers face six bands in 2026 to 2027, with a 42% higher rate starting at £43,663 and a 48% top rate above £125,140. A lumpy withdrawal crosses more thresholds in Scotland, so the case for spreading is arithmetically stronger there and the figures in this piece understate it.

The examples also assume the withdrawal is the only moving part. They ignore the state pension, rental income, dividends, salary sacrifice, gift aid, and the effect of adjusted net income on child benefit or student loan repayments. Any of those changes the numbers. They also assume the pot is uncrystallised, so 25% of each withdrawal is tax free. Money already in drawdown has had its tax free cash taken, and every pound out is taxable.

What the arithmetic cannot tell you is whether the expense is worth funding this way at all. The Financial Conduct Authority's retirement income market data for 2024/25 shows £70,876m withdrawn from pensions across 961,575 plans accessed for the first time, up 35.9% on the year before. That tells you how much money moves. It doesn't tell you how much of it moved at the right time, and no published dataset does.

What would change the conclusion

If the personal allowance were made transferable across tax years, the entire mechanism here collapses. The saving exists only because an unused allowance expires. A carry forward rule for allowances, of the sort HMRC already operates for unused annual allowance on pension contributions, would make the timing question close to irrelevant for anyone below the higher rate threshold.

If the taper were removed and replaced by a straight rate rise, the £6,271 in the salary example largely goes with it. The 60% band is the single largest distortion in the piece, and it is a quirk of drafting rather than a designed rate.

If your marginal rate is going up rather than down, the arithmetic inverts. Everything above assumes the years either side of the expense are low income years. Someone about to inherit, sell a business or start drawing a defined benefit pension may find that this year is the cheap one, and that waiting costs rather than saves.

Withdrawal sequencing is not a strategy so much as a constraint you either notice or don't. The thing worth watching isn't the size of the withdrawal. It's your adjusted net income for the year, and specifically its distance from £100,000, because that is where the marginal cost of the next pound jumps furthest. LedgerTouch shows the running total across accounts; a spreadsheet updated each April shows it once. What neither can do is tell you what the bands look like in 2032.