South Africa Contractor vs Employee Checker

The decision you make before you make the hire

Most international employers approach South African talent the same way. They find someone excellent, agree a monthly figure, send over a one page contractor agreement borrowed from a template site, and get on with the work. For a great many arrangements that is genuinely fine. The trouble is that nobody checks, and the version of the arrangement that ends up running is rarely the version described in the agreement. The person works your hours. They sit in your standups. They use your systems, follow your process, and earn every rand of their income from you. On paper they are a supplier. In practice they are staff, and South African law cares far more about the practice than the paper.

This is not a South African peculiarity. Most legal systems treat classification as a question of substance. What makes South Africa worth understanding properly is that the tests are unusually well documented. The Labour Relations Act sets out a statutory presumption with seven named factors. SARS publishes a twenty indicator grid, weighted into three tiers, describing exactly how its auditors reach a view. The Labour Appeal Court has reduced the common law to three primary criteria. You do not have to guess at how the authorities think, because they have written it down, and this tool simply puts their own tests in front of you.

The reason to run the check before you hire rather than after is that classification is almost entirely within your control at the start and almost entirely outside it later. At the point of writing the role you can decide whether you are buying somebody's time or somebody's output, whether you will direct the method or only the outcome, and whether the person will be free to serve other clients. Once the relationship has run for a year, those facts are established, and no amount of redrafting changes what has already happened.

Three tests, three answers, and why they disagree

The single most common mistake is treating classification as one question with one answer. It is at least three questions, decided by different authorities under different statutes, and they can land in different places for the same person.

The first is a tax question. Does the money you pay count as remuneration under the Fourth Schedule to the Income Tax Act, so that employees tax has to be withheld? The second is a labour law question. Is this person an employee for the purposes of the Labour Relations Act and the Basic Conditions of Employment Act, with the leave, notice, working time and unfair dismissal protections that follow? The third is the common law question underneath both, which asks whether you acquired the person's capacity to work or bought the result of it.

SARS is explicit that these do not have to agree. Interpretation Note 17 works through an example of a person who is an independent contractor under the common law test but is deemed by the statutory test not to be carrying on a trade independently, with the consequence that employees tax is deducted while the person's independent status at common law remains unaffected. That is a genuinely odd result if you expect one answer, and completely coherent once you accept that the tax net and the labour law net are drawn differently and for different reasons. The checker above therefore reports all three separately rather than collapsing them into a single verdict.

Test one: the two SARS statutory tests

Employees tax is only triggered by remuneration, and remuneration specifically excludes amounts paid for services rendered in the course of a trade carried on independently. That exclusion is where genuine contractors sit. Sitting on top of it are two statutory tests which SARS describes as conclusive in nature, meaning they override the common law analysis rather than feeding into it.

The first test has two parts and both must be present. The services must be required to be performed mainly at the premises of the person paying or of the person to whom the services are rendered, where SARS reads mainly as a quantitative measure of more than fifty percent. And the person must be subject to the control or supervision of any other person as to the manner in which their duties are performed, or as to their hours of work. If both parts are satisfied, the person is deemed not to be carrying on a trade independently and the payment is remuneration.

Two details in that test matter enormously for remote hiring. The first is that the control limb is about the right of control, not the exercise of it. Interpretation Note 17 states that it is sufficient if the right of control is contractually present, even if it is not exercised in practice, and gives the example of a business that cannot practically supervise a specialised professional but retains the right to. Writing a supervision clause you never intend to use still satisfies the test. The second detail cuts the other way and is genuinely good news for cross border hiring: a South African working from their own home for an overseas company is not working mainly at your premises. That first limb fails, which means the first statutory test cannot be met, which means the classification falls to the common law grid rather than being decided against you automatically.

The second statutory test runs the other way and overrides the first. A person who employs three or more full-time employees throughout the year of assessment, none of them connected persons such as relatives, is deemed to be carrying on a trade independently. This is the clean answer for hiring a genuine small agency or studio rather than an individual. If the person you are paying runs a real business with real staff, the tax analysis stops there.

Test two: the section 200A presumption

Section 200A of the Labour Relations Act, mirrored word for word in section 83A of the Basic Conditions of Employment Act, lists seven factors. If any one of them is present, the person is presumed to be an employee regardless of the form of the contract. The seven are: the manner of work is subject to another person's control or direction; the hours of work are subject to that control or direction; the person forms part of the organisation; the person has worked for the other person for an average of at least forty hours a month over the last three months; the person is economically dependent on the other person; the person is provided with tools of trade or work equipment; and the person works for or renders services to only one person.

Read that list against a typical full-time remote hire and the problem becomes obvious. Forty hours a month is ten hours a week. Almost any serious remote engagement clears it in the first fortnight. Economic dependence is present the moment you are the person's main income. Working for one person only is the default for anyone hired full-time. You do not need all seven, or three, or two. You need one, and most arrangements trip several without anyone intending it.

The presumption is not available at every income level. It applies to people earning below the BCEA earnings threshold, which the Minister sets by determination in the Government Gazette and which rose to R269,600.90 a year, roughly R22,467 a month, with effect from 1 May 2026. That figure sits comfortably above what most administrative, support and mid-level professional roles pay in South Africa, which means the presumption is available to a large majority of the people an international employer is likely to hire. Senior specialists and engineers frequently earn above it, and for them the presumption falls away, but the common law test does not. A person earning well above the threshold can still be found to be an employee on the realities of the relationship.

What the presumption actually does is shift the burden of proof. It does not make the person an employee; it makes them presumptively an employee unless you can show otherwise. In a dispute that difference decides cases. You arrive having to disprove employment rather than the worker having to prove it, and the evidence you would need is evidence about how the work really ran, which by then is largely out of your hands.

Test three: the dominant impression grid

Underneath the statutes sits the common law, and the common law test in South Africa is the dominant impression test. All aspects of the relationship are weighed and a single overall impression is formed. No indicator is conclusive on its own. What makes the test workable rather than vague is that SARS has published its own version of the weighing exercise: a grid of twenty indicators, set out in Annexure C of Interpretation Note 17, sorted into three tiers by how much weight they carry.

The six near-conclusive indicators are the ones that speak most directly to whether you acquired the person's productive capacity. They are control of the manner of working, the payment regime, whether the person must render the service personally, whether they must be present even when there is no work, whether they are bound to you exclusively, and who bears the risk. The four persuasive indicators establish the degree of control you hold over the work environment: instructions and supervision, reporting, training, and control of productive time. The remaining ten are labels, clauses and circumstances that resonate one way or the other, and SARS is candid that these are the ones most susceptible to deceptive contractual manipulation.

The tiering is the useful part. It tells you where to spend your attention. Providing a laptop is a tier three indicator; who decides the method and what you are paying for are tier one. An arrangement can look like contracting across half a dozen cosmetic factors and still read as employment because the six that matter all point the same way. That is why the tool weights the tiers three, two and one rather than counting boxes, and it is also why SARS warns in the same document that the grid is a guide and should not be used as a checklist to determine a certain score. The number the tool shows you is a direction of travel, not a verdict.

Those six rows are the near-conclusive tier restated as plain questions. If you only have five minutes, answer those and you will know most of what the full grid would tell you. The Labour Appeal Court reached a compatible shortlist from the other direction in SITA v CCMA, holding that the enquiry turns on three primary criteria: the employer's right of supervision and control, whether the person forms an integral part of the organisation, and the extent of the person's economic dependence on the employer.

The part international employers get wrong: who withholds

There is a second question hiding behind the tax test, and it trips up almost every first-time cross border employer. Deciding that a payment is remuneration tells you employees tax is due on it. It does not by itself tell you that you are the person who must withhold and pay it over.

Historically, a foreign company with no representative employer in South Africa fell outside the PAYE withholding net, and the South African worker settled their own liability through the provisional tax regime instead. That position narrowed with effect from 22 December 2023, when the obligation was extended to non-resident employers conducting business through a permanent establishment in South Africa, or that have a representative employer there. A foreign employer with neither of those things still falls outside the registration requirement, which is the position most companies hiring a single South African remote worker are in. The direction of travel is towards more registration rather than less, and further refinement of the permanent establishment requirement was flagged in the 2026 Budget, so treat this as a point to confirm rather than assume.

The practical consequence for you is worth stating plainly, because it is genuinely reassuring. If you are a US, UK, Australian or European company with no office, no branch and no representative in South Africa, hiring one person who works from their own home, you are very unlikely to be walking into a South African payroll registration. Your hire handles their own tax as a provisional taxpayer, which is an ordinary and well understood arrangement in South Africa. That is a large part of why direct hiring from South Africa is administratively lighter than most people expect. It is also why the classification question is worth answering properly rather than worrying about vaguely: for most arrangements the honest answer is that the compliance burden is small and clearly defined.

None of this makes the labour law question go away. Section 200A and the Labour Relations Act attach to the employment relationship rather than to the employer's tax registration status, and the same is true of the contribution regimes under the Unemployment Insurance Act. A foreign employer can therefore be outside the PAYE net and still be dealing with a person who is an employee in every sense that matters to the Commission for Conciliation, Mediation and Arbitration. Those are different questions and the tool keeps them apart for exactly that reason.

What it costs to get it wrong

On the tax side, Interpretation Note 17 puts the position bluntly: an employer that has incorrectly determined that a worker is an independent contractor is liable for the employees tax that should have been deducted, as well as the penalties and interest that go with it. The employer has a right to recover the tax from the worker, which is a right that is often worth less in practice than it sounds, particularly if the relationship has already ended badly. Note also that SARS will not settle the question for you in advance. It expressly declines to issue advance rulings or private opinions on whether a person is an independent contractor, on the basis that only the employer has full knowledge of the relevant facts. The determination is your responsibility and yours alone.

On the labour side the exposure is different in kind. A person found to be an employee acquires the full set of protections retrospectively. That means leave that was never accrued, notice that was never given, working time entitlements that were never observed, and above all the right to refer an unfair dismissal dispute when you end the arrangement. Ending a contract for services is a commercial act. Ending an employment relationship without a fair reason and a fair process is a dismissal, and the two look identical right up to the point where somebody files a referral.

Neither of these is a reason to avoid hiring in South Africa. They are a reason to spend twenty minutes deciding what the relationship is before it starts. The cost of getting it right is a clear head about how you want the work to run and an agreement that says so. The cost of getting it wrong is retrospective and uncapped.

How to structure it deliberately, either way

If you want a genuine contractor relationship, the changes that matter are the six near-conclusive ones and nothing else will substitute for them. Pay for defined deliverables rather than for hours present. Brief the outcome and leave the method alone. Let the person work for other clients and do not write an exclusivity clause. Allow delegation or subcontracting, because a person who can put someone else on the work is plainly running a business rather than holding a job. Push some genuine risk across, so that rework is at their cost rather than yours. And keep the engagement bounded by a result or a term rather than running open ended and indefinitely renewed.

If you read that list and think none of it describes what you want, that is a useful and honest answer. Most companies hiring a full-time remote assistant, coordinator, bookkeeper or developer want exactly the things that make somebody an employee: set hours, your process, your tools, your priorities, and their full attention. In that case the right move is not to disguise it. It is to set it up as employment from the start, with a proper employment agreement, a clear notice period, leave that accrues, and a payroll route agreed before the first payment. Done deliberately, that is a simpler and cheaper arrangement to run than a contractor relationship you would struggle to defend, and it is also a better proposition for the person you are hiring, which shows up in how long they stay.

Whichever route you choose, write it down before the first day. Our contract generator produces an agreement you can adapt, the offer letter generator covers the employment route, and the salary calculator gives you the band to put in it. A written agreement that matches how the work genuinely runs is the single most useful document you can hold if the classification is ever questioned, and it is worth almost nothing if it describes an arrangement that never existed.

How to use the checker

Start with the statutory tests at the top. Three of the four questions have obvious answers for a remote hire: the person is a South African tax resident, they are not working at your premises, and they do not employ three or more staff of their own. The question worth thinking about is the control one, and the right way to answer it is to ask whether you would object if they worked whatever hours they liked in whatever way they liked, provided the output arrived. If the honest answer is that you would object, you hold the right of control whether or not you have ever used it.

Then tick the section 200A factors honestly rather than optimistically. The tool counts them, but the count is mostly for your own benefit, because one is enough. Finally work the grid. Leave an indicator as neither if it genuinely does not apply to your setup rather than guessing, since an unanswered indicator simply drops out of the weighting instead of pulling the result in a direction the facts do not support. If you want to see how the tool behaves on a real arrangement first, load the typical full-time remote hire example at the top, which fills in the pattern most people describe when they say they are hiring a contractor full-time. It is not a flattering result, and that is the point.

Everything runs in your browser and nothing is stored or sent anywhere, so you can put real details in. This is general information built directly from published South African statutes and SARS guidance rather than legal or tax advice, and no classification is ever finally settled by a tool. For a specific engagement, particularly a senior one or one that has already been running for a while, take your completed answers to a South African labour or tax adviser. You will get a faster and cheaper conversation for having done the groundwork.

Sources

Every test in this tool comes from a published statute, code or revenue guidance note. The links below go to the source itself so you can check the wording rather than taking ours for it.

• Labour Relations Act 66 of 1995Section 200A, the seven factor rebuttable presumption of employment
• Basic Conditions of Employment Act 75 of 1997Section 83A, the identical presumption, and section 6(3), the earnings threshold power
• Code of Good Practice: Who is an EmployeeIssued under section 200A(4), sets out how the presumption and the dominant impression test are applied
• SARS Interpretation Note 17 (Issue 5)The two statutory tests, and the twenty indicator dominant impression grid in Annexure C
• SITA (Pty) Ltd v CCMA and Others [2008] ZALAC 1Labour Appeal Court on the three primary criteria and the reality of the relationship
• BCEA earnings threshold from 1 May 2026R269,600.90 a year, set by Ministerial Determination in the Government Gazette
• PAYE obligations for foreign employersThe 22 December 2023 change for non-resident employers with a South African permanent establishment
• Foreign employers and SARS registrationSecond reading of the same amendment, including the representative employer test
• Unemployment Insurance Act 63 of 2001UIF contribution obligations that attach once a relationship is employment
• SARS guidance for employersRegistration, monthly declarations and the practical PAYE mechanics