Tax Loss Harvesting for Retirees: The 0% Bracket Strategy Most Advisors Miss
Tax loss harvesting for retirees is fundamentally different from tax loss harvesting for working professionals, and most articles on the subject get this wrong by treating retirees as simply lower-income versions of the same investor. They aren't. A 38-year-old engineer with a $200,000 salary and a $300,000 portfolio is trying to reduce the tax on income she's earning right now. A 68-year-old retiree with a $90,000 portfolio income and a $1.4 million portfolio is trying to do something completely different — permanently eliminate the embedded tax liability sitting inside positions she's held for thirty years. The mechanics that solve the first problem are not the mechanics that solve the second. The single most powerful tool available to retirees is the 0% long-term capital gains bracket, and the second most powerful is the matched pair — selling a gain and a loss simultaneously to realize the gain at zero tax. Used together over five to ten years, these two techniques can erase six-figure embedded tax liabilities without the retiree ever writing a check to the IRS.
Most retirees know roughly what their account balances are. Very few know what their cost basis is. Even fewer know that the gap between those two numbers — the embedded gain — is a future tax bill they're carrying around silently, growing every year as markets rise. A retiree with a $1.4 million portfolio and a $400,000 cost basis has $1 million in embedded gains. At a 15% federal long-term rate plus 5% state, that's $200,000 in eventual tax. If those positions are held until death, current law gives the heirs a stepped-up basis and the entire tax liability evaporates. But if the retiree needs to sell during their lifetime — to fund medical care, downsize a home, help a grandchild — that tax bill comes due. The point of tax loss harvesting in retirement is to shrink the embedded gain over time, year by year, while paying as close to zero tax as possible.
Why the 0% Bracket Is the Whole Game
For 2026, single filers pay 0% federal tax on long-term capital gains up to $49,450 in taxable income. Married couples filing jointly get the same treatment up to $98,900 in taxable income. After the 2026 standard deduction of $32,200 for couples, that means a married retiree couple can have up to $131,100 in gross income — including realized long-term gains — and pay zero federal tax on the gains portion.
Most retirees fall comfortably below that threshold. Social Security checks for an average couple run $30,000 to $50,000 combined. A 4% safe withdrawal from a $1 million IRA produces $40,000 in ordinary income. Add a small pension and a few thousand in dividends, and many retirees sit at $80,000 to $100,000 of total income — meaning they have $30,000 to $50,000 of headroom inside the 0% bracket each year, every year, that they can fill with realized long-term gains and pay nothing for it.
This is not a loophole. It is the federal tax code working exactly as designed. Long-term capital gains for taxpayers in the lower two ordinary income brackets (10% and 12%) are taxed at 0%. Most retirees living on Social Security and modest withdrawals are in those brackets. The 0% LTCG rate is the explicit policy result.
What makes the 0% bracket so powerful in combination with tax loss harvesting is that filling it doesn't reduce future opportunities — it permanently raises basis. Every dollar of gain realized at 0% becomes a dollar of new basis when the position is repurchased the same day. The portfolio's market value doesn't change. But the embedded tax liability shrinks by the harvested amount, multiplied by the future rate the heir or surviving spouse would have paid.
How Bracket Filling Actually Works
Worked example #1. Margaret and Tom are 67 and 65, married, retired in 2024. Their 2026 income looks like this:
- Combined Social Security: $48,000
- Tom's pension: $18,000
- IRA RMDs (Margaret's only, Tom isn't yet 73): $14,000
- Dividends from taxable brokerage: $9,000
That totals $89,000. Of the Social Security, roughly 85% is taxable at the federal level given their other income, so effective AGI is about $81,200. After the standard deduction of $32,200, taxable income before any realized gains is $49,000.
Their headroom inside the 0% long-term capital gains bracket: $98,900 − $49,000 = $49,900.
They own a Vanguard Total Stock Market ETF position purchased in 2010 at $60 per share, now worth $260 per share. They own 600 shares for a total value of $156,000 and a cost basis of $36,000. Embedded gain: $120,000.
In 2026, they sell shares totaling exactly $49,900 in realized long-term gain. At $200 per share gain on a $260 sale price, that's 250 shares sold at $260 each — proceeds of $65,000 — generating a gain of $50,000 (close enough; the precise math would call for 249.5 shares but brokerages sell whole shares). They pay $0 in federal tax on those gains because the entire amount fits inside their 0% bracket.
They immediately repurchase 250 shares of the same ETF at $260 per share. (No wash sale issue applies — wash sale rules only disallow losses, not gains.) Their cost basis on those 250 shares is now $260 instead of $60. They've raised the cost basis on 250 shares by $200 per share, eliminating $50,000 of embedded gain at a cost of zero federal tax.
Repeated annually, this technique unwinds Margaret and Tom's $120,000 embedded gain over roughly two and a half years, after which the entire 600-share position has a basis near current market value. If markets rise during that period, they continue raising basis on whatever embedded gain remains. Over five years, an aggressive bracket-filling program can eliminate hundreds of thousands of dollars of future tax liability for a retiree couple — paid for entirely by the structure of the federal tax code.
What If You're Above the 0% Bracket?
Many retirees, particularly those with pensions, large RMDs, or substantial non-retirement investment income, sit above the 0% threshold. They aren't done — they just need a different tool. This is where matched pairs come in.
A matched pair is the simultaneous realization of a gain and a loss, sized so that the loss fully offsets the gain. The investor sells a position with an unrealized gain and a position with an unrealized loss in the same tax year, in roughly equal amounts. The two cancel out for tax purposes. The investor immediately repurchases both — the gain position retains its higher basis (raised), and the loss position is replaced with a correlated but not substantially identical security to avoid the wash sale rule.
Worked example #2. Robert is 71, widowed, single filer. His 2026 income includes Social Security ($32,000), a $50,000 pension, RMDs of $40,000, and $20,000 in taxable interest and dividends. His effective AGI is approximately $128,000. After the standard deduction of $16,100 for single filers, his taxable income is $111,900 — well above the single-filer 0% LTCG threshold of $49,450, putting him in the 15% bracket on any realized gains.
He owns a position in an S&P 500 index fund purchased in 2008 with $200,000 in unrealized gain. He also owns an emerging markets ETF, purchased in 2021 near a peak, currently sitting at a $35,000 unrealized loss. He realizes both simultaneously: $35,000 gain on the S&P fund, $35,000 loss on the EM fund. Net taxable gain: zero. Federal tax owed: zero.
He immediately repurchases an S&P 500 ETF (basis raised by $35,000), and replaces the EM fund with a different international fund — say, a developed-markets ex-US fund — that maintains roughly similar exposure without triggering wash sale issues. His portfolio looks essentially identical the next morning. His embedded tax liability is $35,000 lower than it was the day before.
Done annually for ten years, the matched pair strategy can erase $350,000 of embedded gain in Robert's portfolio at zero net federal tax. Even if some years don't have a perfectly sized loss available, partial offsets — a $25,000 loss against a $40,000 gain, leaving $15,000 taxable at 15% for $2,250 — are still dramatically better than realizing the gains without any offset.
What Most Articles About Tax Loss Harvesting for Retirees Get Wrong
The standard advice goes something like this: "Tax loss harvesting is less valuable in retirement because retirees are typically in lower tax brackets, so the savings per dollar harvested are smaller."
This is true if you measure tax loss harvesting only by its ability to offset ordinary income or generate tax savings on realized gains in the current year. It misses the entire point of harvesting in retirement, which is not income tax reduction but embedded liability reduction. A retiree in the 0% LTCG bracket isn't saving 15 cents per dollar of harvested loss — they're saving zero cents on the current loss but using that loss (or the bracket headroom) to permanently raise basis on dollars that would otherwise have been taxed at 15-20% (or worse, with NIIT, 23.8%) sometime in the future, possibly during their lifetime, possibly during a surviving spouse's reduced-bracket years.
The right way to measure the value of harvesting in retirement is the present value of future tax avoided. A 67-year-old who eliminates $50,000 of embedded gain this year, on a position they likely would have sold in the next ten years to fund late-life expenses, has avoided roughly $7,500 to $10,000 of future tax. That's the actual savings. Done annually, the compounding effect on a retiree's lifetime tax burden is substantial.
The other thing the standard advice gets wrong is the assumption that retirees should "let it all step up at death." This works only if the retiree never sells anything during their lifetime — never funds long-term care, never helps a child with a down payment, never moves to a smaller home, never has a medical event that requires liquidity. In practice, retirees sell. The question is whether they sell with embedded gains intact or with basis already raised. The latter outcome requires planning a decade in advance.
The Sequencing Problem: When to Harvest, When to Defer
Retirement income planning interacts with TLH in ways that aren't obvious. RMDs are forced sales of pre-tax retirement assets that show up as ordinary income. Social Security taxation depends on a complicated formula involving "provisional income" that can push more of a retiree's benefits into the taxable category if other income rises. IRMAA Medicare premium surcharges kick in at specific income thresholds and can add hundreds of dollars per month to a retiree's costs.
This means that filling the 0% LTCG bracket up to the line is not always the right answer. A retiree who realizes $30,000 of additional gain to fill the bracket may inadvertently cross an IRMAA threshold and trigger $2,000 of additional Medicare premiums. Or they may push more of their Social Security into the taxable category, which has the same downstream effect. Or they may push themselves out of a state property tax exemption tied to AGI.
The right approach is to model all of these thresholds simultaneously and harvest only up to the point where the next dollar of realized gain triggers a downstream cost. For some retirees this means harvesting $30,000 of gains in a year where the bracket allows $50,000. For others it means realizing exactly $0 in gains some years and $80,000 in others, depending on Roth conversion plans, RMD timing, and irregular income events. This is exactly the kind of multi-variable optimization that becomes intractable for a human to do annually with a spreadsheet — and tractable for software that scans the lot-level position continuously and models the threshold landscape.
Connecting It to the Broader Picture
For retirees with significant embedded gains, the combined strategy looks like this: in years when the 0% bracket has headroom, fill it with intentional gain realization. In years when income pushes above the 0% threshold, use matched pairs to neutralize realized gains against harvested losses. In years when neither tool is fully available, harvest losses anyway and bank them as carryforward for future years when they can offset gains. Over a 20-year retirement, the cumulative reduction in embedded tax liability often exceeds $200,000 for a household with a seven-figure taxable portfolio — entirely without changing how the portfolio is invested.
The decisions are individually small. The cumulative effect is not.
For background on how the rate brackets actually work, see capital gains tax rates 2026. For investors still close to the NIIT threshold, our net investment income tax NIIT 2026 piece explains why every harvested dollar above $250,000 in MAGI is worth more than it appears. The lot-level mechanics that make this work — and the reasons most brokerage default methods produce inferior outcomes — are covered in detail in our optimal tax lot selection deep dive. And for a real-world walkthrough of how lot-level harvesting saves five figures on a single position, see Sarah's $31,200 NVDA case study.
The 0% bracket is one of the most generous features of the federal tax code, and almost the only people positioned to use it well are retirees. The strategy isn't aggressive, it isn't novel, and it isn't a loophole. It's just arithmetic — applied consistently, year after year, until the embedded liability that most retirees never knew they were carrying has been quietly eliminated.