Key takeaways
- Hargreaves Lansdown's best-buy table on 6 August 2026 paid £8,040 a year for life on a £100,000 pot at 65 — single life, level, no guarantee.
- Taking that same £8,040 from a £100,000 drawdown pot empties it at 83.4 on a 5% net return, and at 89.8 on 7%.
- ONS's 2024-based projections give a man turning 65 this year a 61.8% chance of reaching 83, and a woman a 72.4% chance.
- Die at a man's cohort life expectancy of 20.17 further years and the annuity's implied return is 5.75%, against 5.67% on a 20-year gilt.
- Of the 88,430 annuities sold in the 2024/25 year, 80.3% were level-only and 38.2% went to the provider the buyer already had.
£8,040 a year for life, or a pot that has to earn 8.74% to match it
Annuity or drawdown? Here's the arithmetic that decides it — and then the reason the arithmetic isn't really the point.
On 6 August 2026, Hargreaves Lansdown's best-buy table paid £8,040 a year on a £100,000 pension at age 65. That's a single-life, level annuity with no guarantee period. It works out at an annuity rate of 8.04%.
Keep the same £100,000 in drawdown and take £8,040 a year out of it, and the pot has to earn 8.74% a year, net of every charge, just to stand still. Below that it drains. The only question is when it hits zero.
So the break-even isn't a rate of return. It's an age. Live past it and the annuity paid more. Die before it and your estate keeps whatever the drawdown pot still holds. The rest of this piece is about where that age sits, and how sure anyone can be of reaching it.
Change the return assumption and the break-even age moves by 20.6 years
The chart plots the age at which a £100,000 drawdown pot runs dry while paying out £8,040 a year. Four assumptions hold it up, and all four are stated rather than hidden. Withdrawals come out at the start of each year. The income stays level in cash terms, to match a level annuity. The return is net of platform and fund charges. And nothing is indexed on either side.
With no growth at all, the pot lasts to 77.4. At 3% net it reaches 80.2. At 5% net, 83.4. At 7% net, 89.8. At 8% net it stretches to 98.0. That's a spread of 20.6 years, generated entirely by an assumption nobody can verify in advance.
Two things follow. The first is that anyone quoting a single break-even age is quoting their own return forecast without saying so. The second is that the drawdown side of this comparison is a distribution and the annuity side is a point. A real portfolio doesn't deliver 5% every year — it delivers a sequence, and a bad early sequence empties the pot sooner than the same average would suggest. Our guide to sequence risk in retirement withdrawals works through why order matters when money is coming out.
A break-even age means nothing until you attach a survival probability
ONS's 2024-based projections put cohort life expectancy at 65 in 2026 at 20.17 further years for a man and 22.86 for a woman. Cohort life expectancy allows for the mortality improvements expected across the rest of that person's life. The more familiar period figure doesn't, which is why period tables understate how long today's 65-year-olds tend to live.
An average is still an average. What matters here is the whole survival curve, and ONS publishes the single-year mortality rates you need to build it. Chaining those rates forward, one age and one calendar year at a time, gives the survival path for the cohort turning 65 this year. The reconstruction reproduces ONS's own published cohort life expectancy to within a tenth of a year, which is a reasonable check that the chaining is right.
On that basis a man turning 65 in 2026 has a 71.4% chance of being alive at 80, 61.8% at 83, and 33.6% at 90. For a woman the same three numbers are 80.1%, 72.4% and 46.3%.
Line those up against the break-even ages. On a 5% net return the annuity wins from 83.4, and roughly 6 in 10 men and 7 in 10 women get there. On a 7% net return it wins from 89.8, and only a third of men do. On a 3% net return it wins from 80.2, which most of both sexes reach. The return assumption doesn't just move the age. It moves the odds from likely to unlikely.
At average longevity the implied return barely beats a gilt
Treat the annuity as an investment for a moment. You hand over £100,000 and receive £8,040 a year until you die. The internal rate of return depends entirely on the death date.
Die at a man's cohort life expectancy of 20.17 years and the implied return is 5.75% a year. At a woman's 22.86 years it's 6.57%. Live to 95 and it's 7.74%.
Now the comparison that reframes the whole decision. On 5 August 2026 the Bank of England's 20-year nominal zero-coupon gilt yield was 5.67% and the 10-year was 4.96%. At average longevity, then, the annuity's implied return sits a fraction above the long gilt and 0.79 points above the 10-year. That is roughly what the insurer is invested in.
The extra is the mortality credit — the money released to survivors by the people in the pool who die early. At average longevity it's worth almost nothing. At 95 it's most of the return. An annuity isn't priced to beat markets on average. It's priced to pay you when you've outlived the table, which is a different product entirely.
One buys longevity insurance, the other keeps optionality — and only one of those is a risk transfer
Strip the arithmetic away and the two options insure different things.
An annuity transfers longevity risk to an insurer. It removes the possibility of reaching 95 with nothing left, and it does so by removing the possibility of a windfall too. That's what insurance does. Nobody expects a positive expected return from buildings cover.
Drawdown transfers nothing. It keeps the pot, the market exposure, the flexibility to spend unevenly and the residual value at death. It also keeps the risk that a long life and a poor sequence arrive together. Every flexible-withdrawal framework is an attempt to manage that risk without transferring it — see withdrawal guardrails and flexible spending rates and the international evidence on safe withdrawal rates outside the US, which is far less comforting than the US-only version.
The closest familiar comparison isn't a fund. It's a state pension top-up, which is also a lump sum exchanged for indexed income for life — the arithmetic of what a voluntary Class 3 National Insurance year buys runs on exactly the same logic, at a much smaller scale.
The level annuity is an inflation bet, and 80.3% of buyers take it
Here's the strongest objection to everything above. The £8,040 is fixed forever in cash terms. Every year of inflation takes a bite out of it, and the break-even arithmetic above quietly assumes the drawdown withdrawal is equally unindexed.
The inflation-proofed version is available and it's expensive. On the same Hargreaves Lansdown table, a single-life RPI-linked annuity with a 5-year guarantee started at £5,402 — 32.8% below the level rate. That's the price of the protection, paid entirely up front in foregone income, and it pushes the early years of the comparison heavily towards drawdown.
Buyers overwhelmingly decline it. FCA data for the 2024/25 year shows 80.3% of the 88,430 annuities sold were level-only. A reasonable reading is that most buyers want maximum income now. A less comfortable reading is that a level annuity looks like a hedge against longevity while leaving inflation entirely unhedged, and that the two risks compound in exactly the scenario the annuity is bought for — a very long life. This piece does not compute a break-even for the RPI-linked contract, because that needs an inflation path, and any inflation path here would be an assumption dressed as a finding.
Most people don't annuitise, and most who do never see the best rate
The FCA's retirement income market data for the year to March 2025 counted 88,430 annuity sales against 349,992 new drawdown policies, out of 961,575 pots accessed for the first time. The pots differ too: the money used for annuities averaged £76,641 per sale, against £152,486 per new drawdown policy.
Rate dispersion is the part with a price tag. Legal & General's published table, correct as at July 2026, showed £7,296 a year at 65 on a £100,000 pot for a single-life annuity with a 10-year guaranteed minimum payment period. Hargreaves Lansdown's best-buy row with a 5-year guarantee was £7,977, and with no guarantee £8,040. The terms aren't identical, so the gap isn't purely shopping around — but on the closest comparable rows it's 9.3%, for life, on an irreversible purchase.
In the same FCA data, 38.2% of annuities were bought from the provider the customer already held a pot with. Only 29.2% of annuity purchases used regulated advice, and 53.2% used neither advice nor Pension Wise guidance. That last figure is worth sitting with, because buying a lifetime annuity is not a decision you can revisit. Across all retirement products, the FCA recorded 30.6% of first-time pot access as advised.
The inheritance argument for drawdown gets smaller on 6 April 2027
The most cited reason to stay in drawdown is that the pot passes on. That's true, and it's about to be worth less.
HMRC's technical note, updated 29 May 2026, is explicit: "From 6 April 2027, most unused pension funds and pension death benefits will be brought within the value of a deceased person's estate for Inheritance Tax purposes." The Finance Act 2026 received Royal Assent on 18 March 2026 and applies the change to deaths on or after that date.
Income tax sits on top. The same note states that "If the member was over age 75, all death benefits are taxable" — at the beneficiary's marginal rate. A drawdown pot inherited from someone who died after 75 can therefore face inheritance tax on the estate and income tax on the beneficiary. The residual value that made drawdown look generous to heirs is not the headline number.
On the other side, gov.uk confirms you can usually take up to 25% of a pension as a tax-free lump sum, capped at £268,275, and that applies whichever route you take with the remainder. It isn't a point of difference between the two options.
The limitations of this comparison, stated plainly
These are single quotes from a single week. Annuity rates move with gilt yields, and the £8,040 headline is a best-buy for a healthy person at an average postcode. Your own quote could be higher — enhanced rates apply where health or lifestyle shortens expected life — or lower.
The break-even model uses a constant net return. Real portfolios don't do that, and the constant-return version flatters drawdown at the long end, because it never lands a poor decade early. The model also ignores tax, which is defensible only because annuity income and drawdown withdrawals are both taxed as income at the same marginal rate. It ignores partial annuitisation, which is neither option and is what a lot of people actually do.
The mortality data cannot tell you when you personally die. ONS projections are projections, and the further out they run, the less reliable the assumed improvements become — ONS says so itself. Population averages also ignore the strong link between wealth and longevity, and the people with £100,000 pots are not a random sample of the population. If anything, that pushes the survival probabilities above the figures used here.
One more limitation, and it's the biggest. This compares a £100,000 pot in isolation. It ignores the state pension already underneath it, other income, a partner's position and how much of your spending is genuinely fixed.
What would change the conclusion
If gilt yields fell sharply, annuity rates would follow them down, the £8,040 would shrink, and every break-even age in the chart would fall with it. The 8.04% rate on offer in August 2026 is a function of where yields sit now, not a permanent feature.
If you'd qualify for an enhanced rate, the whole calculation shifts. A health-impaired quote raises the income and lowers the break-even age at the same time, while your personal survival curve sits below the population one. Those two effects push in opposite directions, and only an actual quote resolves them.
If the level-versus-indexed choice were made differently, the answer changes shape rather than degree. A 32.8% lower starting income buys a contract that keeps its purchasing power, and its break-even against drawdown lands much later in nominal terms.
The honest summary is that this decision is an insurance question wearing an investment question's clothes. The break-even age is computable to one decimal place. The probability of reaching it is a population statistic that says nothing about you. What's left is a preference: whether you'd rather hold the risk of outliving your money, or pay an insurer a mortality credit worth roughly a gilt yield at average longevity to take it away.
If the drawdown route is the one being weighed, the number to keep an eye on isn't the break-even age at all. It's the withdrawal rate against a pot that moves every day, which LedgerTouch tracks continuously. The annuity route replaces that number with a certainty, which is precisely what it's selling.