Tax-Free Lump Sum vs Drawdown: the Bill to Age 95

Key takeaways

• The full new State Pension is £241.30 a week in 2026/27, which is £12,547.60 a year. That leaves £22.40 of the £12,570 personal allowance for every other pound of income.
• On a £400,000 pot drawn as a level real income to age 95, taking the 25% up front and phasing it both produced an income tax bill of £87,308 over the 29 years.
• Cashing the same pot in one go at 66 costs £126,849 of income tax in that single year, £39,541 more than either phased route.
• The 25% tax-free component was worth £29,146 across those 29 years at this pot size, and the amount doesn't change with the order you take it in.
• Planning to 95 rather than 85 cuts the sustainable level withdrawal from £26,987 to £20,101 a year, a reduction of 25.5%.

Tax-free lump sum vs drawdown: over a full horizon the tax bill is the same

You're 66, you have a defined contribution pot, and one person has told you to take the 25% tax-free cash before the rules change. Another has told you to leave it invested and draw it gradually. You want to know which one costs less tax.

Run both to age 95 and the honest answer is neither. On a £400,000 pot drawn as a level real income for 29 years, taking £100,000 tax free at 66 and drawing the remaining £300,000 as taxable income produced a total income tax bill of £87,308. Taking every withdrawal as an uncrystallised funds pension lump sum, 25% tax free and 75% taxable, produced £87,308. It's the same number, because it's the same arithmetic in different clothes.

What the choice does move is everything around the tax bill: what happens when one year's income breaks a band, where the money sits once it has left the pension, what happens to your annual allowance if you're still contributing, and who pays inheritance tax on whatever is left. Those are the parts of the tax-free lump sum vs drawdown question that carry real money.

The State Pension leaves £22.40 of your personal allowance

Start with the figure that shapes everything downstream. The Department for Work and Pensions publishes the full rate of the new State Pension for 2026/27 as £241.30 a week. Across 52 weeks that is £12,547.60. GOV.UK gives the standard personal allowance as £12,570 for the tax year running from 6 April 2026 to 5 April 2027.

The headroom is £22.40. For someone on the full new State Pension and nothing else, the personal allowance is already spent. Every pound of taxable pension income after that is taxed, and it starts at 20%.

The basic rate band runs to £50,270 of income. Subtract the State Pension and £37,722.40 of taxable pension income fits inside the 20% band before the 40% rate begins. Because only 75% of an uncrystallised funds pension lump sum is taxable, that corresponds to a gross withdrawal of about £50,297 in a single tax year. Above roughly that level, the marginal rate on the taxable slice doubles.

This is why the order of withdrawal is a tax question at all. Spread across years, income sits low in the bands. Compressed into one year, it climbs them.

The model: FCA projection rates, a level real income, and 29 years

The numbers here come from one arithmetic model, stated in full so you can disagree with it precisely.

The growth assumption is the FCA's own. COBS 13 Annex 2 sets the intermediate nominal projection rate for a personal pension at 5%, with an intermediate price inflation assumption of 2%. Compounding those against each other gives a real return of 2.9412% a year, and every figure below is in today's money at that rate.

The pot is £400,000 at age 66. The withdrawal is level in real terms, paid at the start of each year, and it exhausts the pot with the payment at 94, so the money runs to age 95. At 2.9412% real over 29 payments the annuity factor is 19.90, which makes the sustainable withdrawal £20,101 a year in today's money.

Two assumptions matter more than the growth rate. The first is that the personal allowance and the band thresholds rise with prices over the whole period, which is what working in today's money implies and is not what has happened recently. The second is that the tax-free money, once outside the pension, is sheltered from tax on its own income and gains. Neither holds automatically, and both get their own section below.

The order of withdrawal only matters when one year breaks a band

Here is the comparison, all four routes measured the same way: total income tax paid on the pension across the 29 years to age 95, including tax on the State Pension.

The chart above plots those four totals side by side, and the shape of it is the finding: two bars of exactly equal height, one much taller, one in between.

Rows two and three are identical to the pound. Taking a £100,000 pension commencement lump sum at 66 leaves £300,000 of taxable pot, which at the same annuity factor pays £15,076 a year. Phasing pays £20,101 a year of which 75%, again £15,076, is taxable. The tax-free entitlement is 25% of the money either way, and 25% of a fixed pot is a fixed amount whenever you take it.

Row one is the expensive one. Encashing the whole pot means £100,000 tax free and £300,000 added to the State Pension in a single year, giving taxable income of £312,547.60. The personal allowance tapers away entirely above £125,140, at £1 for every £2 of income over £100,000. Applying the 20%, 40% and 45% bands to that leaves £126,849 of income tax, an effective rate of 42.3% on the taxable £300,000. Against either phased route, encashment costs £39,541 more, and it does so in year one rather than spread over three decades.

Row four is the control. If none of the pot were tax free, the same £20,101 of annual spending would be fully taxable and the bill over 29 years would be £116,454.

The 25% is worth £29,146 here, and it doesn't get bigger by being taken sooner

Subtract row three from row four and the tax-free component is worth £29,146 to this retiree over 29 years. That is 20% of the £145,730 that passes through the tax-free slice, because at this pot size the whole of the taxable income sits inside the basic rate band. For someone whose other income pushed them into the 40% band, the same entitlement would be worth twice as much.

That asymmetry is not a rounding artefact of the model. It's the design. The ISA vs pension comparison turns on the same mechanism, running in the opposite direction at the contribution end.

What the £29,146 does not do is grow because you took it early. The lump sum allowance caps the tax-free amount at £268,275, and GOV.UK states the rule plainly: "You can usually take up to 25% of the amount built up in any pension as a tax-free lump sum. The most you can take is £268,275." A separate lump sum and death benefit allowance of £1,073,100 covers serious ill-health and certain death benefit payments. Below the cap, 25% of a pot is 25% of a pot.

The one genuine difference sits in what the 25% is measured against. Money left uncrystallised keeps earning growth that is itself 25% tax free when eventually drawn. Money taken as a lump sum at 66 has fixed its tax-free entitlement at the pot's value on that day. Our model neutralises that by assuming the extracted lump sum grows at the same rate in a shelter, which is the assumption the next section takes apart.

Once the lump sum leaves the pension it has to sit somewhere

A pension is a tax wrapper. Inside it, investment income and gains are not taxed. Take £100,000 out at 66 and that shelter is gone for that £100,000, and it has to be rebuilt.

The ISA subscription limit for the 2026 to 2027 tax year is £20,000. Moving £100,000 into ISAs at that rate takes five tax years, during which the unsheltered balance is exposed to tax on its dividends, interest and gains. That drag does not appear anywhere in the table above, because the table assumes it away.

Cutting the other way is inheritance tax. Most unused pension funds and pension death benefits come within the value of an estate for inheritance tax purposes from 6 April 2027. HMRC's own assessment expects around 213,000 estates with inheritable pension wealth in 2027 to 2028, of which 10,500 face an inheritance tax liability where previously they would not, and roughly 38,500 pay more. It puts the average increase at around £34,000 per affected estate. That reform removes the argument, widely made before it, for leaving a pension untouched purely to pass it on. Where an estate is near the thresholds, the interaction with the UK inheritance tax thresholds and the residence nil rate band taper does more work than the withdrawal order does.

Taking taxable income triggers the money purchase annual allowance

This is the one hard asymmetry between the two routes, and it only bites if you're still paying into a pension.

The annual allowance is £60,000 for the current tax year. HMRC's Pensions Tax Manual sets the money purchase annual allowance at £10,000 for 2023 onwards, and it applies "for the tax year in which they first flexibly access their benefits and every subsequent tax year". The manual's list of trigger events is built around payments: a trigger occurs immediately before the first payment from a newly designated flexi-access drawdown fund, and when an uncrystallised funds pension lump sum is first paid.

So the two routes are not symmetric for someone aged 66 who is still working and still contributing. A phased UFPLS strategy takes taxable income by construction, and takes it in year one. A pension commencement lump sum without any accompanying drawdown income does not appear on that payment-triggered list. The gap between £60,000 and £10,000 of allowance is £50,000 a year of contribution capacity, which for a higher earner can exceed the entire tax difference in the table. It's the one place in the tax-free lump sum vs drawdown comparison where the two routes genuinely do not tie.

Running to 95 rather than 85 costs a quarter of the annual income

Age 95 is not a forecast. It's a margin, and the margin has a price.

The ONS 2022-based cohort life tables put it clearly: "people aged 65 years in the UK in 2023 can expect to live on average a further 19.8 years for males and 22.5 years for females". That's an average age at death of about 84.8 for men and 87.5 for women. Planning to 95 means funding 10.2 years beyond the male average and 7.5 beyond the female one.

Run the same £400,000 pot to 85 instead, at the same 2.9412% real return, and the sustainable level withdrawal is £26,987 a year rather than £20,101. Stretching the plan by ten years costs £6,886 a year of spending, a cut of 25.5%. The tax bill falls with it, to £4,044 a year across the shorter horizon, but that's small consolation: you're paying more tax per year on more income and simply stopping sooner.

That is the actual trade in decumulation, and it's not a tax trade. It's the choice between spending less now and having something left at 92. Someone who guesses short and lives long is left with the State Pension of £12,547.60 a year. Someone who guesses long and dies at 80 leaves an estate that, from April 2027, is inside the inheritance tax net. Neither error is symmetric with the other, which is why the annuity vs drawdown comparison keeps reappearing in this part of a retirement plan.

The order of returns matters as much as the order of withdrawals here. A poor first decade compounds against a fixed withdrawal in a way an average return never shows, which is the whole of sequence risk, and it is invisible in a level-return model like this one.

The counter-case: the 25% was never designed to be neutral

The strongest objection to reading any of this as a settled arithmetic problem is that the rule itself is contested and openly criticised by the people who model it.

The Institute for Fiscal Studies, in its February 2023 report A blueprint for a better tax treatment of pensions, listed three problems with the tax-free component. One of them is directly relevant here: it "favours taking at least 25% of the pension in the form of a lump sum, or a series of lump sums, rather than buying an annuity or entering formal drawdown without taking a 25% lump sum first". A tax rule that changes behaviour is a rule under pressure.

The IFS also sized it. Abolishing the tax-free component would raise, for contributions made in 2022-23 and measured over the saver's lifetime, "an additional tax burden of £5.5 billion", of which "just under 70% of the total £5.5 billion tax increase would fall on the top 20% of earners". They floated capping it at 25% of the first £400,000 of pension wealth, which on the wealth distribution then approaching retirement "would affect only about one-in-five retirees".

None of that is a prediction. It is evidence that the £29,146 in this model is a policy variable, not a constant, and that the direction of pressure on it has been downward for some years.

Where this model breaks

Four limits, stated plainly.

The return is a straight line. Real portfolios do not deliver 2.9412% every year, and a level real withdrawal against a volatile portfolio is a different problem, as the safe withdrawal rate literature shows.

The thresholds are assumed to track prices. If the personal allowance and the £50,270 band boundary are held flat in cash terms while incomes rise, more of every withdrawal is taxed and the encashment penalty in row one grows. Working in today's money hides that entirely.

The pot size drives the result. At £400,000, all the phased taxable income sits in the 20% band, which is exactly why rows two and three tie. Raise the pot far enough that annual withdrawals cross £50,297 gross and the routes stop matching, because a front-loaded lump sum changes which years cross the band.

And this is one household with one income source. Add rental income, a defined benefit pension, or a spouse with unused personal allowance, and the band arithmetic changes before the withdrawal order does. This model does not include the scale of what people actually do: the FCA recorded 961,575 pension plans accessed for the first time in 2024/25, 349,992 drawdown sales, 88,430 annuity sales, and £70,876m withdrawn in the year. That is a very wide distribution of circumstances behind one median case. Where the money is needed all at once rather than as income, withdrawal sequencing across tax years changes the bill more than the choice of product does. Where the money is needed all at once rather than as income, withdrawal sequencing across tax years changes the bill more than the choice of product does.

What would change the conclusion

Three things, in order of how likely they are to move the number.

A change to the tax-free component. The lump sum allowance is £268,275 today. A cap set lower, or a percentage below 25%, would break the tie between rows two and three immediately, because the entitlement would stop being a fixed fraction of the pot.

The band thresholds. The whole reason encashment costs £39,541 more is that it crosses into 40% and 45% and loses the personal allowance. Move the boundaries far enough and that penalty shrinks or grows without anyone changing their behaviour.

And the inheritance tax reform landing on 6 April 2027. Once unused pension funds sit inside the estate, the question stops being what the withdrawal costs you and starts being what the residual costs the people who inherit it. That is a different calculation from the one in the table above, and for estates over the thresholds it's likely to be the larger one. If you're tracking a drawdown pot alongside ISAs and taxable accounts, LedgerTouch shows the composition, though the tax arithmetic is still yours to do.

All figures are for the 2026/27 UK tax year and were checked on 5 September 2026. Tax rules change, and this one is under active review.