Everything that changed for 2026
The indexed figures that rose, the state rates that moved mid-year and were backdated, and the thresholds that have not been adjusted since 1978.

The changes that affect almost everyone
The standard deduction rose to $16,100 single and $32,200 joint for 2026, from $15,200 and $30,400. That is $900 more sheltered from tax for a single filer — worth $198 at a 22.00% marginal rate, before anything else changed.
Bracket thresholds moved with inflation too. The 10.00% band now runs to $12,400 for a single filer, from $11,925; the 12.00% band to $50,400, from $48,475; the 22.00% band to $105,700, from $103,350.
Those adjustments exist to stop inflation quietly raising real tax rates, and they mean identical nominal income costs slightly less tax this year than last. Someone whose salary did not change at all sees a small reduction, which is the indexation working as designed.
The Social Security wage base rose to $184,500, so the 6.20% now applies to more of a high earner's wages before it stops. Rates themselves are unchanged at 6.20% and 1.45%, as they have been for decades.
The Child Tax Credit is $2,200 per qualifying child with up to $1,700 refundable. The Earned Income Tax Credit maximum reached $8,231 for three or more children, $7,316 for two, $4,427 for one and $664 for none.
The figures that did not change, and why that is a change
Not everything is indexed, and the thresholds that are not have a quiet cumulative effect that nobody legislates and nobody announces.
The Social Security taxation thresholds — $25,000 and $34,000 single, $32,000 and $44,000 joint — have never been indexed. They were set in 1983 and 1993. When they were written, a small minority of beneficiaries paid tax on benefits; that is no longer true, and no decision was taken to make it so.
The net investment income tax thresholds of $200,000 and $250,000 have not moved since the tax took effect in 2013. The $200,000 additional Medicare tax threshold is the same, and the Child Tax Credit phase-out starts are fixed in nominal terms too.
The $3,000 annual limit on deducting capital losses against ordinary income was set in 1978 and would be roughly $15,000 today if it had been indexed. The $25,000 rental passive-loss allowance dates from 1986. The $400 self-employment tax threshold has been unchanged since 1990.
Each year those unmoved figures reach slightly further down the income distribution. It is the least visible tax change there is, it happens without a vote, and it is why "nothing changed for me this year" is rarely quite true.
What moved at state level, which is where most errors are
State legislatures work on their own calendars and frequently backdate, which makes state figures the ones most likely to be wrong in any published table. Of 37 states we checked against their own department of revenue, 12 carried a wrong rate, threshold or credit in the compiled sources everyone uses.
Georgia cut its rate under HB 463 with effect from 1 January, enacted in May — so a table published in April was correct when written and wrong by summer. South Carolina's structure was altered by H.4216. Both are the same failure mode: legislation passed after the compilation and applied retroactively.
California stopped fully taxing military retirement, excluding $20,000 for 2025 through 2029 — a change that compiled sources were still showing the old treatment for. Oregon's earned income credit was published at 17% in sources we checked when the statute says 9%, which nearly doubles the stated benefit.
Mississippi, Hawaii, Arizona, Massachusetts, Maryland, Louisiana, New Jersey and Ohio all carried corrections of their own. Ohio's was the largest in effect: a source treating a flat rate as flat from the first dollar overstated the bill by 43%, because it ignored the band below which nothing is taxed.
The direction of travel across the country remains toward flat rates. 13 states now apply a single rate and seven of those moved from graduated brackets since 2021, usually with further reductions written into the statute on a schedule — which means the rate that is correct today is scheduled to be wrong.
How to tell whether what you are reading is current
The single fastest check on any tax page is the standard deduction. For 2026 it is $16,100 single and $32,200 joint. A page showing $15,200 or $30,400 is a year behind, and everything computed from it is wrong.
We found that exact error on sites ranking on the first page for "2026 income tax calculator" when we audited the category in August. Several linked to the correct IRS page from the same screen, which suggests the figures were entered once and never revisited rather than sourced incorrectly.
For state figures, check the date. Not a copyright year in the footer — a date attached to the tax figures themselves saying when they were last verified and against what document. Very few pages have one, and its absence is the strongest available signal about a state rate.
Be particularly careful with anything published between January and June, because that is when state legislatures pass changes that apply retroactively to 1 January. A spring table can be accurate on the day it is written and wrong by the time you read it, through no fault of whoever wrote it.
Everything on this site carries the document it came from and the date it was checked. Where a figure has not yet been read off a primary source, the page says so instead of implying it has — because "we have not looked" and "it does not exist" are different claims, and only one of them is a reason to stop looking.
What the changes are actually worth
Indexation is not a tax cut, but it does mean unchanged nominal income costs slightly less than last year. On $85,000 single, the $900 increase in the standard deduction alone is worth $198 at a 22.00% marginal rate.
The bracket adjustments add to that, and their effect is largest for anyone whose income sits just above a threshold. Moving a threshold upward moves a slice of income from a higher rate to a lower one, which is worth the rate difference on that slice.
If your salary rose by roughly the rate of inflation, the two effects broadly cancel and your effective rate is about the same as last year. That is indexation working as intended: it stops inflation raising real tax rates without anyone legislating it.
If your salary did not rise, you are slightly better off. If it rose faster than inflation, your effective rate rose — not because rates changed, but because more of your income now sits in higher brackets, which is progressivity rather than a change in policy.
Where the arithmetic genuinely shifts is around the thresholds that did not move. A household whose income crossed one of the unindexed thresholds this year faces a new charge or a phasing-out credit that nothing announced, and the cause is inflation rather than any decision.
The limits that move each year alongside the brackets
Retirement and health account limits are adjusted annually on their own schedules, and they are the changes with the most direct planning consequences because they define how much you are allowed to shelter.
Workplace retirement plan limits — 401(k), 403(b) and similar — rise with inflation in set increments, with additional catch-up amounts available from a certain age. IRA limits move separately and more slowly, as do the income ranges over which a traditional IRA contribution stops being deductible and a Roth contribution stops being permitted.
Health savings account limits are adjusted annually for individual and family coverage, with their own catch-up provision. The HSA remains the only account that avoids income tax, Social Security and Medicare when funded through payroll — which makes it the most tax-efficient dollar available to most employees.
Flexible spending account limits and the amount that may be carried over move too, and unlike an HSA the money is largely use-it-or-lose-it, which makes the annual election a decision rather than a default.
Confirm each against the IRS notice for the year before setting a contribution rate. These are the figures most often quoted from memory, they change more often than brackets, and setting a payroll deduction from a stale number either wastes headroom or triggers a correction.
Why so many published tables are a year behind
Federal figures are published once, in a single revenue procedure, in the autumn before the year they apply to. They are easy to source correctly and easy to enter once and forget — which is exactly the failure we found: pages linking to the correct IRS document while displaying the previous year's numbers.
State figures are the harder problem. Fifty-one jurisdictions, fifty-one legislative calendars, no common publication date and no single document. Keeping them current is not a once-a-year task, which is why almost everyone treats it as one.
Retroactive enactment makes it worse. A state that passes a rate change in May with effect from 1 January invalidates every table published in the first five months of the year, and nothing about those tables signals that they have been overtaken.
Scheduled future reductions add a third layer. Several states have written multi-year rate reductions into statute, with steps that sometimes depend on revenue triggers being met. A rate that is correct today is scheduled to be wrong, and whether the step happens may not be known until close to the date.
Which is why the useful discipline is per-figure dating rather than per-page updating. A page refreshed last week can still contain a state rate that was entered eighteen months ago, and only a date attached to the figure itself reveals it.
Planning around figures that are scheduled to move
Several states have written multi-year rate reductions into statute, stepping down on a schedule and in some cases conditional on revenue triggers being met. That produces a specific planning question: whether to accelerate or defer income across a step.
Where a rate is scheduled to fall next year, deferring income into that year is worth the rate difference. Where a trigger might not be met, the step may not happen — so the decision is made under genuine uncertainty rather than against a known future rate.
Federal provisions with expiry dates create the same problem at a larger scale. Anything described as applying "for tax years 2025 through 2029" is a temporary measure, and planning that assumes it continues is planning on an expectation rather than on the law.
The practical response is to prefer decisions that are good under both outcomes, and to be explicit when a decision is a bet on a scheduled change. Accelerating a deduction into a year you know the rate is high is safe; deferring income into a year whose rate depends on a revenue trigger is not.
And to re-check in January rather than assuming. Of 37 states we checked against their own department of revenue, 12 carried a wrong figure in the compiled sources — and scheduled changes that did or did not happen are one of the reasons.
The federal rules that stayed exactly the same
The seven bracket rates are unchanged: 10.00%, 12.00%, 22.00%, 24.00%, 32.00%, 35.00%, 37.00%. Only the thresholds moved. This is the most misreported aspect of annual indexation, because a headline saying "brackets changed" is technically right and reads as though the rates did.
FICA rates are unchanged at 6.20% and 1.45%, as they have been for decades. Only the wage base moved, to $184,500.
The capital gains rate structure of 0.00%, 15.00% and 20.00% is unchanged; the income thresholds at which each applies moved with inflation.
The net investment income tax remains 3.80% above $200,000 single and $250,000 joint — the same thresholds as in 2013, because this one has never been indexed.
And the mechanics did not change: brackets are still slices, credits still beat deductions of the same size, FICA is still charged from the first dollar with no deduction, and a refund is still your own money coming back. Nothing in the structure of the calculation moved, which is why last year's understanding remains correct even where last year's figures do not.
Where the figures in this guide come from
Every number above comes from IRS Rev. Proc. 2025-32, § 3.01, Tables 1-4 and each state's own department of revenue publication, read off the document itself rather than off a summary of it. The prior-year comparisons come from the competitor audit of August 2026, which contrasted what several published calculators were serving against the official figures one by one.
That distinction is not pedantry. When we audited this category in August 2026, sites ranking on the first page for "2026 income tax calculator" were publishing a standard deduction of $15,200 single — the previous year's figure — while linking to the correct IRS page from the same screen.
State figures were read one state at a time off each department of revenue's own publication. Of 37 states reviewed, 21 matched the compiled sources everyone uses and 12 did not. The errors were overwhelmingly about timing rather than structure: rates superseded by legislation passed after the compilation, several of them backdated to 1 January.
So each figure on this site carries the document it came from and the date it was checked, and where something has not yet been read off a primary source the page says so rather than implying it has. "We have not looked" and "it does not exist" are different claims, and only one of them is a reason to stop looking.
The check we would suggest running on anything you read about this year's tax changes, here included: find the underlying figure, and compare it against the source it claims to come from. It takes two minutes and it settles the question that no amount of confident writing can.
How the adjustment is actually computed
The figures do not rise by whatever inflation was. They are computed from a specified price index measured over a specified twelve-month period ending in the autumn before the tax year, then rounded — which is why the increase in a given year rarely matches the inflation figure in the news.
The rounding matters more than it sounds. Different provisions round to different increments — some to the nearest $50, some to $100, some further — so two figures adjusted by the same percentage can move by different proportions, and a threshold can stay unchanged in a low-inflation year while another rises.
The index used was changed in 2017 to one that generally rises more slowly than the one used before, on the reasoning that consumers substitute between goods as prices change. The practical effect is that thresholds climb slightly more slowly than they would have, which compounds quietly over years.
The lag is the other structural feature. Because the measurement period ends the autumn before, a year of unusual inflation shows up in the following year's figures rather than in the current ones — so the adjustment is always describing conditions that have already passed.
None of which is a criticism; it is simply why "the standard deduction went up by inflation" is an approximation. The exact figures come from the revenue procedure, and the exact figures are what a calculation has to use.
When to check what, through the year
Autumn, before the year begins. The IRS revenue procedure for the following year is published, and with it the brackets, standard deduction and credit amounts. This is also when contribution limits for the next year are announced, which is what you need before setting a payroll election in January.
January. State changes take effect, including any that will later be enacted retroactively to this date. Your first payslip of the year reflects the new withholding tables, which is the moment to check whether the W-4 on file still matches your circumstances.
Spring. State legislatures pass the changes that will be backdated. Anything you read between January and June about state rates is provisional, and this is the period in which published tables are most likely to be overtaken without being marked as such.
Mid-year. The best moment to check withholding, because there is still half a year to correct anything and the year-to-date figures are large enough to be meaningful. A shortfall found in June is spread over thirteen paychecks; the same shortfall in April is one payment.
Autumn again. Anything you want to move between tax years — a bonus, a charitable contribution, a capital gain, a Roth conversion — has to be decided before 31 December, and the decision needs next year's figures, which have just been published.
Where to go next
Questions
- What is the 2026 standard deduction?
- $16,100 single, $32,200 married filing jointly, $24,150 head of household — up from $15,200 and $30,400. Being 65 or over, or blind, adds $1,650 per condition per person.
- Did tax brackets change for 2026?
- The rates did not — they remain 10.00%, 12.00%, 22.00%, 24.00%, 32.00%, 35.00%, 37.00%. The thresholds moved with inflation, so identical nominal income falls slightly lower in the structure than last year. For a single filer the 10.00% band now runs to $12,400.
- Why is my tax lower on the same salary?
- Inflation indexation. The standard deduction and every bracket threshold rise each January, so unchanged nominal income is taxed slightly less. It is designed to stop inflation raising real tax rates by stealth, and it works — for the figures that are indexed.
- Which tax thresholds are not adjusted for inflation?
- Several significant ones. The Social Security taxation thresholds date from 1983 and 1993, the net investment income tax and additional Medicare thresholds from 2013, the $3,000 capital loss limit from 1978, the $25,000 rental passive-loss allowance from 1986, and the $400 self-employment threshold from 1990. Each reaches further down the income distribution every year.
- What changed at state level?
- More than most tables show. Georgia cut its rate retroactively to 1 January under legislation passed in May, South Carolina altered its structure, and California stopped fully taxing military retirement. Of 37 states we checked, 12 carried a wrong figure in compiled sources.
- How do I check whether a tax page is current?
- Find the standard deduction it uses and compare it against $16,100 single for 2026. If it shows $15,200, it is a year behind. Then look for a date attached to the figures themselves rather than to the page — its absence is the strongest signal there is.