How gross-to-net pay is actually calculated

Almost every explanation of net pay you will find says the same thing: take your gross salary, subtract social contributions, subtract income tax, and what is left is your take-home pay. That is roughly right, and it is also missing the part that actually matters. The interesting question is not what gets subtracted but in what order, because the order decides how much income tax you owe in the first place.

Countries do not agree on the order. Of the 105 countries in our dataset, 70 calculate income tax on your salary after social contributions have been taken out. The other 35, including the United States, the United Kingdom, Canada, Australia, Norway, Czechia and Hungary, calculate income tax on the full gross figure and treat social contributions as a separate, parallel deduction.

The four steps

Stripped to its bones, a payroll calculation anywhere in the world runs through the same four steps:

1. Start from gross. This is the number in your employment contract. It is not what your employer pays; that figure is higher and is covered in what an employee actually costs.
2. Take employee social contributions. Pension, health, unemployment insurance, depending on the country. These are usually a flat percentage of gross, sometimes capped above a ceiling.
3. Work out the taxable base. Subtract the tax-free allowance. In 70 countries you also subtract the social contributions from step 2 here. In the other 35 you do not.
4. Apply the income tax rates to that base, flat or bracket by bracket, and subtract the result.

Step 3 is the whole game. Everything else is arithmetic that any two calculators will agree on.

Why the order costs you real money

Take a deliberately simple hypothetical: gross pay of 3,000 a month, employee social contributions of 10%, a tax-free allowance of 500 a month, and a flat 20% income tax.

Identical rates, identical allowance, 60 a month of difference. Over a year that is 720, and the gap widens as the social contribution rate rises. This is why comparing two countries by their headline income tax rates tells you very little: a 20% rate charged on gross is not the same tax as a 20% rate charged after contributions.

A real worked example

Estonia is the cleanest country to trace by hand, because it is the only country in our dataset with a genuinely flat income tax and no bracket structure to keep track of. The 2026 figures: 22% income tax, a tax-free allowance of roughly 8,400 a year, and two employee contributions, unemployment insurance at 1.6% and a funded pension at 2%. Both of those reduce the taxable base.

You can check that against the Estonia salary calculator and get the same 1,658. The reason the effective rate lands almost five points below the headline rate is entirely the allowance, which is the subject of its own guide.

What this model deliberately leaves out

Any calculator, ours included, models a standard case. The things that most often push a real payslip away from the estimate are:

• Contribution ceilings. Many countries stop charging pension or health contributions above a threshold, so the effective rate falls at higher salaries rather than staying flat.
• Allowances that taper. Lithuania and the United Kingdom, among others, shrink the tax-free amount as income rises, which creates a band where the marginal rate is much higher than any published bracket.
• Local and regional taxes. Municipal income tax in the Nordics, regional variation in Spain, state and city tax in the United States. A national-level model cannot know your address.
• Household circumstances. France taxes the household, not the individual. Joint filing in Germany and the United States changes the outcome substantially for couples.

None of these make a national-level estimate useless. They do mean the estimate is a starting point for a conversation with your employer or accountant, not a substitute for your actual payslip. The specific ways the two drift apart are worth knowing, and they are collected in why your payslip never matches an online calculator.