The same salary, two filing statuses: what marriage actually changes on a paycheck
Published 2026-11-09 · updated 2026-11-09 · figures for tax year 2026
Marriage does not change the tax rates you pay. It changes the width of the taxable base you pay them on: the standard deduction doubles, every bracket for a joint return is twice as wide as the single version, and the two incomes are pooled. On one salary the effect is always positive. On two, it depends on how close the two salaries are — and in a handful of state schedules it goes the other way entirely.
The federal change is arithmetic, not policy
For 2026, the single standard deduction is $16,100 and the joint figure is $32,200 — exactly double. The brackets follow the same rule: the 10% band, the 12% band, the 22% band, all twice as wide for a joint return. Put a single salary against that wider base and less of it reaches each rate than it would on a single return.
The same 90,000 in three states
| State | Single take-home | Joint take-home | Joint is ahead by |
|---|---|---|---|
| California | $66,994 | $73,850 | $6,856 ($571/mo) |
| Texas | $72,145 | $76,675 | $4,530 ($378/mo) |
| New York | $67,875 | $73,014 | $5,139 ($428/mo) |
The federal half of that difference is identical in all three states, because federal rules do not vary by state. What varies is the state half: a state with no income tax can only contribute the federal gain, while a progressive state contributes its own as well. The measured spread runs from $4,530 to $6,856 a year on the same gross salary.
Where the joint return actually saves: the federal base
Look at the federal tax line alone rather than at take-home, and the mechanism is plain. On $90,000 the single filer pays $10,970 of federal income tax; the joint filer pays $6,440. The gap is $4,530, produced entirely by pushing $16,100 more of the salary below the tax floor and widening the bands above it. Social Security and Medicare do not move at all: they are charged per worker on wages, not per household, so the FICA line is $6,885 on both returns.
The case people miss: two earners, not one
The $90,000 example above is a single earner with a non-earning spouse, which is the case where joint filing is unambiguously better. Two earners of $45,000 each are a different situation, and the useful comparison is not "joint vs single" on the household total but "joint vs what the two payrolls actually withheld".
| Arrangement | Gross | Federal tax | California tax |
|---|---|---|---|
| Two jobs of 45,000, each withheld as single | $90,000 | $6,440 | $1,497 |
| One 90,000 return, filed jointly | $90,000 | $6,440 | $1,655 |
The federal columns agree — $6,440 withheld against $6,440 owed — because two single returns each receive a $16,100 deduction, and $32,200 is exactly what one joint return receives. The state column does not agree: California brackets for a joint return are not simply twice the single ones at every band, so merging the income moves $158 of state tax either way. That asymmetry, not the federal schedule, is where the phrase "marriage penalty" comes from.
How to use this on a paycheck
A joint return is a filing status for the year, not a switch on a payslip. What changes immediately is the W-4: both spouses file as married, and then exactly one of them should claim the full joint deduction — the other claims nothing, or the two split it through the worksheet. Claiming it twice produces the same under-withholding described in the two-jobs case: each payroll believes the household has $32,200 of untaxed income when the household has it once.
Filing separately is available and occasionally worth it — typically where one spouse has large medical expenses or where income-based student loan repayments are involved — but it is a return-level decision. On a paycheck, it looks like a single filer with a married person's withholding, which is the worst of both.
Put a number on it
The same mechanisms on an actual salary. Every figure here is computed by the engine that serves the calculators, so it matches the page it opens — nothing is copied.