K53 pass rate 2026: why so many people fail the new learners test

Fewer people are passing the learner’s test than a few years ago. That much is on the record. What it means is a more careful question than the headlines suggest, and the useful answer turns out to have nothing to do with the statistics at all.

What the numbers say

Two figures are worth knowing, and both need their source attached to them.

Nationally: the RTMC, through its spokesperson Simon Zwane, in May 2026, said the learner’s licence pass rate has fallen from 68% to 40%. That is a government body’s own figure, carried by several news publishers. We have not seen the data underneath it.

In the Western Cape: figures from the Freedom Front Plus, reported by MyBroadband in July 2026, put the pass rate at about 17% across roughly 142 100 tests since the new system launched in that province in May 2025, against about 63% before. Three things about that number matter as much as the number: it is one province, it originates with a political party, and it reaches us through a news site rather than from the department that holds the data.

You will see that Western Cape figure turned around and printed as a national headline about how many people fail. That number is not national. Subtracting a provincial pass rate to produce a scarier national fail rate is not a statistic, it is a marketing decision, and it is one you should distrust wherever you see it.

What the numbers do not say

They do not say the test got harder. We would like to be able to tell you that, because it is a tidy story. We cannot, because at least three other things changed at the same time and each one would push a pass rate down on its own:

• Questions are drawn at random. Two candidates side by side are not writing the same paper.
• The draw reaches further into the syllabus. The Western Cape’s Acting Director for Transport Administration and Licensing told EWN in December 2025 that the old printed booklets did not use all the available questions, and that the computerised test does.
• Whatever used to circulate stopped working. If a printed paper could be seen in advance, a randomised draw ends that.

A number that moved while three things changed cannot tell you which of the three moved it. What the evidence does support is narrower, and more useful to you: far fewer people are passing it without preparing properly.

So why do people fail?

Here the honest answer is better than the statistical one, because it is about mechanics you can do something about.

1. They study one test. They are marked on three.

This is the big one. The pass is not a percentage; it is three thresholds joined by the word “and” in the 2006 notice: 22 of 28 for rules, 23 of 28 for signs, 6 of 8 for controls. You clear all three or you fail.

So a person can get every rules question and every signs question right (56 out of 56), drop three of the eight controls questions, and go home having scored 61 out of 64. Ninety-five percent. Fail.

Nothing about that is unusual or unlucky. It is how the test is written, and it means “I’m getting about 80% overall” is a measurement of nothing the test does. The three sections goes through what is in each one and where the marks go.

2. The smallest section carries the most risk

Eight controls questions, and you may get two wrong. That is the whole margin, and it is the section people give the least time to, because it looks small and because a car’s controls feel obvious once you have sat in one.

For a light motor vehicle the Department of Transport’s manual numbers eleven controls. Eleven things. It is an evening’s work, and it protects a quarter of your risk.

3. Wording beats people more often than pictures

This one comes from first-hand accounts rather than data, and we will label it as such. Candidates and instructors consistently describe the traps as verbal, not visual: the difference between must and may, questions asking for the “most correct” answer, items that ask you to select a combination of numbered statements, hand signals where the direction the officer is facing is not stated, and pairs of signs that differ in one detail.

IOL reported applicant complaints about question wording in May 2026, which is at least a second, independent hint in the same direction. It is not a statistic and we do not present it as one. Treat it as a warning about how to read a question, slowly and twice.

4. Nobody is allowed to tell them what they got wrong

This is the mechanism that turns one failure into two.

The regulations say the examiner shall only give the results of the test to the applicant and may not discuss the questions and answers of the test. So a candidate walks out with three numbers and nothing else: 21 out of 28 for signs, say, and no way to know which signs, or whether it was one topic missed seven times or seven topics missed once.

Then they book again, study exactly as they studied before, because they have no information that would tell them to do anything differently, and the second attempt looks a great deal like the first. Breaking that loop is mostly a matter of reading the three numbers you do get, properly.

5. “About 80%” feels much safer than it is

The pass marks work out to roughly 79%, 82% and 75%, so being “about 80% good” sounds like enough. It is not, and the reason is the random draw.

Your questions are a sample of what you know, so each section is a spot-check. Being 80% solid across the board means each of three spot-checks is close to a coin flip, and you need all three to land. On our own modelling, someone 80% solid would clear all three sections about 3 times in 10. At 90% it is about 9 times in 10.

Our own estimate, from the section sizes and pass marks in Government Notice R.93 of 2006. It models each section as a random draw on what you know. It is a projection, not an official statistic, and it is not a prediction about you.

The practical consequence: aim well above the pass mark in every section, and measure yourself section by section rather than as one average. That is how PassPath scores you, and it is the only reason we can tell you which of the three would have failed you.

The figures at the top of this guide are kept on their own page, with every publisher and date against them and a row for each province: the K53 pass rate tracker. Eight of the nine provinces say unknown there, because nothing has been published for them.