Methodology
The formulas behind every calculator on this site, the assumptions each one makes, and the things they deliberately do not model.
A calculator that won't show its working is asking to be trusted for no reason. So here is the working: how each tool computes what it computes, which IRS document every figure came from, and the part most sites leave out, which is what each model knowingly gets wrong.
Every IRS figure lives in one file, with its citation attached
No number is typed into a page. Every IRS figure on this site sits in a single data file next to the notice or revenue procedure it came from, so when the IRS publishes new limits each autumn, one file changes and every page showing a limit changes in the same build.
That arrangement isn't a convenience. It's there because the most common failure in this category is a page still displaying a figure that was correct two years ago, and the cause is nearly always the same: the §402(g) deferral limit got typed into forty places and updated in thirty-eight.
The build refuses to publish a figure from a data block that hasn't been marked as verified against its primary source. If a number can't be confirmed, the tool that needs it switches itself off. It says why, on the page. A calculator showing a plausible guess at an IRS limit is worse than one showing nothing, because the guess gets acted on.
| Figure | 2026 | Source |
|---|---|---|
| Employee deferral limit | $24,500 | IRC §402(g) |
| Catch-up, age 50+ | $8,000 | IRC §414(v) |
| Enlarged catch-up, ages 60–63 | $11,250 | SECURE 2.0 §109 |
| Prior-year wages above which the catch-up must be Roth | $150,000 | IRC §414(v)(7)(A) |
| Total annual additions | $72,000 | IRC §415(c) |
| Compensation limit | $360,000 | IRC §401(a)(17) |
| Highly compensated employee threshold | $160,000 | IRC §414(q) |
| IRA contribution limit | $7,500 | IRC §219(b) |
| IRA catch-up, age 50+ | $1,100 | IRC §219(b)(5)(B) |
| Roth IRA phase-out, single | $153,000 to $168,000 | IRC §408A(c)(3) |
| Roth IRA phase-out, married filing jointly | $242,000 to $252,000 | IRC §408A(c)(3) |
| Roth IRA phase-out, married filing separately | $0 to $10,000 | IRC §408A(c)(3)(B) |
| IRA deduction phase-out, single and covered by a plan | $81,000 to $91,000 | IRC §219(g) |
| IRA deduction phase-out, filing jointly and covered | $129,000 to $149,000 | IRC §219(g) |
| IRA deduction phase-out, only your spouse is covered | $242,000 to $252,000 | IRC §219(g)(7) |
| IRA deduction phase-out, filing separately and living together | $0 to $10,000 | IRC §219(g)(4) |
| Qualified charitable distribution limit | $111,000 | IRC §408(d)(8) |
| Federal income tax brackets | 7 bands | Rev. Proc. 2025-32 |
| Standard deduction | $16,100 single, $32,200 jointly, $24,150 head of household | IRC §63(c); Rev. Proc. 2025-32 |
| Additional standard deduction, age 65 or over or blind | $1,650, rising to $2,050 if unmarried and not a surviving spouse | IRC §63(f); Rev. Proc. 2025-32 §4.14 |
| Social Security taxable maximum | $184,500 | SSA annual announcement |
| Social Security and Medicare employee rates | 6.2% and 1.45% | IRC §3101 |
| Additional Medicare tax | 0.9% on wages above $200,000 single, $250,000 jointly | IRC §3101(b)(2) |
| Social Security taxation, base amounts | $25,000 single, $32,000 jointly, $0 filing separately while living together | IRC §86(c)(1) |
| Social Security taxation, second-tier amounts | $34,000 single, $44,000 jointly | IRC §86(c)(2) |
| Share of the benefit that can become taxable | up to 50%, then up to 85% | IRC §86(a) |
| Age required minimum distributions start | 73 born 1951–1959, 75 born 1960 or later | SECURE 2.0 §107 |
| Uniform Lifetime Table divisors | ages 72 to 100 | Treas. Reg. §1.401(a)(9)-9; Pub. 590-B Table III |
| Excise tax on a missed distribution | 25%, or 10% if corrected in time | SECURE 2.0 §302 |
| Plan loan cap | $50,000, or 50% of the vested balance | IRC §72(p)(2)(A) |
| Plan loan floor | $10,000 | IRC §72(p)(2)(A)(ii) |
| Plan loan term, other than a residence loan | 5 years | IRC §72(p)(2)(B) |
| Additional tax on an early withdrawal | 10% | IRC §72(t) |
| Withholding on an eligible rollover distribution | 20% | IRC §3405(c) |
| 403(b) 15-year service catch-up | $3,000 a year, $15,000 over a lifetime, $5,000 times years of service | IRC §402(g)(7) |
| Service it requires | 15 years | IRC §402(g)(7) |
| 457(b) final-years catch-up | 2 times the annual limit, in the last 3 years before the plan's retirement age | IRC §457(b)(3) |
| Qualified long-term care distribution cap | $2,600, or 10% of the present value of the vested benefit | IRC §72(t)(2)(N) |
Each year's new money earns half a year of growth, by convention
Contributions are assumed to arrive evenly through the year. Each year's new money therefore earns about half a year of return, and the arithmetic is one line: a balance of B receiving a contribution C at rate r becomes B(1+r) + C(1 + r/2).
That is the whole growth model.
It's a convention, not a law. Someone paid monthly who defers a fixed percentage of salary lands very close to it. Someone who front-loads in January to hit the §402(g) cap by the summer does slightly better than the model shows, and someone who drops a lump sum in on the last payroll of December does slightly worse. Nobody I know of models that difference, and for most people it is smaller than the rounding on their next pay rise.
Tax runs through the real bracket table, never a flat rate
Federal income tax is computed through the seven bands published in Rev. Proc. 2025-32, for the filing status you select, applied to income after the standard deduction and after any pre-tax deferral. You supply no flat rate. An average rate can't tell you the thing most worth knowing, which is whether one more percentage point of contribution drops you out of a bracket.
FICA is Social Security at 6.2% on wages up to the annual taxable maximum, Medicare at 1.45% on all wages, and the additional 0.9% Medicare tax on wages above the statutory threshold. It's computed on full wages. A pre-tax 401(k) deferral cuts your income tax and does nothing at all to your payroll tax, which catches people out constantly, so the calculators say it on the page instead of quietly netting it off.
The standard deduction includes the age-65 addition, and does not include the temporary senior deduction. Income tax is computed after the standard deduction for your filing status, and from 11 September 2026 the engine adds the further deduction IRC §63(f) gives a filer aged 65 or over: $1,650, or $2,050 for someone unmarried and not a surviving spouse, both read from Rev. Proc. 2025-32 §4.14. It matters most in the drawdown tools. A retired filer who qualifies is the normal case there rather than the exception, and leaving the addition out overstated the tax on every withdrawal those tools modelled. The separate temporary senior deduction created by the One Big Beautiful Bill is not applied, and that is deliberate: the IRS release naming it publishes no amount and no phase-out, so there is no primary figure to put in the data file, and a figure that isn't in the data file can't reach a page here. If you are 65 or over, the tools understate your deduction by whatever that turns out to be.
State and local income tax is not modelled anywhere on this site yet. Doing it properly means capturing how each state treats retirement income. Several exempt it outright, several more apply partial exclusions with age or dollar thresholds, some tax Social Security benefits and most don't, and every one of them needs its own citation. That work isn't done. Until it is, the tools say "federal only" instead of applying one national average and calling it close enough.
A Roth comparison that skips the side account is rigged
Deferring C into a traditional 401(k) at marginal rate t reduces take-home by C(1−t) and returns C t in tax. Total outlay is C. That is identical to a Roth contribution of C, so a comparison that leaves the C t sitting uninvested has quietly compared a plan against the same plan minus a cash gift, and it hands you the wrong answer every time.
So a side account exists. The traditional path carries a taxable one holding the invested tax saving, and the Roth path carries nothing beyond the plan itself. Cost basis is tracked, only the gain is taxed on exit at the capital gains rate, and the dividend share of the return carries an annual drag.
One test keeps this honest. With equal tax rates before and after retirement, and a frictionless side account, the Roth and traditional paths must finish exactly level, because multiplication commutes and it makes no difference whether tax comes off before the growth or after it. If the model ever fails that test the model is wrong, and it's the first assertion in our test suite for that reason.
The match is pre-tax in both. Employer contributions normally land in a traditional bucket even when your own deferrals go to Roth, and they're taxed on withdrawal either way.
Match is computed band by band, the way a plan administrator does it
Tiered formulas are worked out one band at a time. The order matters here. Dollar caps are applied after the percentage calculation, because that's the order plans using them apply them, and doing it the other way round overstates the match for anyone above the cap. The optimal contribution rate is found by walking upward in tenths of a percent until the match stops increasing, and it's never reported above the §402(g) ceiling for your age.
Vesting uses the cliff and graded schedules permitted by IRC §411(a)(2). Graded schedules are applied from year two, which is the usual reading. Your plan document governs. Where it disagrees with this site, it wins.
One fixed rate of return is the largest lie in every projection here
Real markets don't deliver an average. The order in which the good and bad years arrive matters more than most people expect, and two portfolios with identical average returns can finish a long way apart depending on when the bad years landed. A probabilistic model is planned. Until it exists, treat every projection here as a sketch of one possible path and not a forecast.
These tools also don't model IRMAA surcharges on Medicare premiums, the net investment income tax, plan-specific fees, or the SECURE 2.0 rule that catch-up contributions must be Roth for higher earners.
Correction, 11 September 2026. The paragraph above used to list required minimum distributions among the things not modelled. That was wrong, and wrong in a way this page contradicted two paragraphs later by naming the RMD calculator as the owner of the Uniform Lifetime Table. Distributions are modelled. The RMD calculator computes them straight from the table, and the monthly income calculator works out from your birth year whether one is due yet, then shows it beside the income you were planning to draw.
What neither tool decides for you is whether to take more than the minimum your 401(k) has to pay out.
Every formula above belongs to a named tool
Each rule above has an owner. The growth and compounding convention drives the projection on the homepage and the contribution planner, and the bracket engine drives the paycheck calculator and the withdrawal calculator. The equal-cost comparison is the Roth versus traditional tool, the match formulas are the employer match calculator, and the IRS Uniform Lifetime Table drives the RMD calculator.
A wrong figure gets corrected on the page and left visible
If you find an error, tell us. A figure gone stale against the IRS notice that set it, or a formula that gives the wrong answer at an edge case, is worth more to us than a compliment. Corrections are made to the page and noted there, never quietly overwritten.
We ran the check on everyone else too. How accurate 401(k) calculators are reports what we found when we tested ten of them against the IRS source, and sets out the method in full so you can run it yourself on any calculator, this one included.