How these numbers are worked out
Pay is decided by published schedules. So every figure here traces to one, and the parts we cannot source are inputs you supply rather than numbers we invent.
Rates verified against the official sources — how we check.
The rates, and where each one comes from
| What | Value for 2026-27 | Source |
|---|---|---|
| Resident tax brackets | Nil to $18,200, then 15c, 30c, 37c and 45c | ATO — Tax rates for Australian residents |
| Medicare levy | 2%, with the single low-income phase-in from $28,011 | ATO — individual income tax rates |
| Low income tax offset | $700, tapering to nil at $66,667 | ATO — individual income tax rates |
| Study and training loans | Nil to $69,528, then the marginal scale to a 10% ceiling | ATO — Study and training support loans rates and repayment thresholds |
| Study loan indexation | 2.8% applied 1 June 2026, the lower of CPI and WPI | ATO — Study and training loan indexation rates |
| Annual leave | 4 weeks of the employee's own ordinary hours; 5 only where an award grants it | Fair Work Ombudsman — Annual leave (Fair Work Act 2009 sections 86–87) |
| Annual leave loading | Award-dependent, not NES; usually the higher of 17.5% and the usual penalties | Fair Work Ombudsman Library — Annual leave loading in awards and agreements |
| Super guarantee | 12%, annual contribution base $270,830 | ATO — Super guarantee (rates and maximum contribution base) |
Two places we follow the table, not the formula
The ATO’s published study loan table says the repayment above $129,717 is $9,028 plus 17c for each dollar over that threshold. Fifteen per cent of the band below it actually comes to $9,028.35, so the ATO has rounded its own base constant down by thirty-five cents. We use the rounded figure, because the ATO’s worked examples use it and our answer should match the notice of assessment you receive.
The same table ends with a flat 10% row starting at $186,051. That is a band, not a ceiling, so there is a small step up in the repayment right at the edge. We reproduce the step rather than smoothing it away, for the same reason. Whether the legislation itself rounds is an open question we are still checking, and this note will change if the answer does.
Study loans: indexation first, repayment second
A study loan balance does not behave like an ordinary debt, and the order of events is where most calculators go wrong. Indexation is applied on 1 June to the part of the loan that has been unpaid for more than 11 months. The extra amount your employer withholds all year is not credited to the loan until your return has been lodged and a compulsory repayment has been worked out — months after that indexation. Our payoff projection runs the cycle in that order, so it does not quietly credit a year of withholding before indexation and report a shorter payoff than you will get.
The projection assumes you are no longer borrowing, and it has to assume future indexation rates and future repayment thresholds, because neither is published in advance. Both are inputs you control on the HECS repayment calculator, and the page says which figures are assumptions rather than published rates.
Leave: the National Employment Standards, not your award
The leave calculators model the NES floor, which every award and agreement must at least match. Two things follow. The entitlement is four weeks of the employee’s own ordinary hours — Fair Work’s published example is a part-timer on 20 hours a week accruing 80 hours a year — so we never assume a 38-hour week, and that example is pinned by a test. And leave loading is not in the NES at all: it defaults to zero, and the loading page explains the “higher of” comparison most awards make rather than asserting 17.5%.
We do not model individual awards, and we will not guess at one. Where the answer depends on your award — the fifth week for shiftworkers, which penalties count towards loading, whether an annualised salary absorbs it — the page says so and points at the award.
Annual liability, not your employer’s withholding
The pay calculator works out the tax you owe for a full year and divides it evenly. Your employer instead uses the ATO withholding tables, which assume every pay period is typical. If your income varies, if you started part-way through the year, or if you hold more than one job, the two will differ during the year and get reconciled at your tax return. That is the honest description, and it is why we do not claim to reproduce a payslip.
What is deliberately not modelled
- The Medicare levy surcharge and the private health insurance rebate.
- Family and senior Medicare levy thresholds, and the seniors and pensioners tax offset.
- Non-resident, working holiday maker and Norfolk Island schedules.
- Salary sacrifice, which needs the concessional cap modelled with it and gets its own page.
- Individual awards and enterprise agreements. The leave pages model the NES floor and the shape of award loading rules; the specific rates, penalties and shift definitions in your award are Fair Work’s to publish, and its own tool is authoritative on them.
- Tax on a termination payout, including unused leave paid on leaving. That has its own rules and is not the same as ordinary withholding.
Anything on that list changes a real answer for the people it applies to. Listing it is more useful than a calculator that quietly assumes none of it exists.
How a wrong rate gets caught
Rates live in dated data files, never in page copy, and each carries its source and the date it was checked. Every engine is pinned by automated tests to the published figures, including the ATO’s and Fair Work’s own worked examples. If a rate is edited and the tests are not, the build fails. That is the design, not an inconvenience: it means a stale number cannot reach the site quietly.
Every schedule is re-verified each 1 July, when the new financial year’s figures are published. If you spot something wrong before we do, please say so.
Who does the checking, and what we are and are not qualified to say, is set out on the about page. This site is independent: no bank, super fund, payroll vendor or lender has any part in it.