How Accurate is My Car's Speedo?

In these increasingly speed-obsessed times, how accurate is your car's speedo, really? What are the rules? And how can you possibly check it?

Free audio.jpg

This question, which I get a lot of, is in pigeon English - my favourite kind. But (if I am interpreting it correctly) he makes a good point.

"Why have I notice lately they Speedo’s in cars a lieing. In my mums brand new xtrail her car seems to be doing 6 Kms less ph then what the gps says in my Audi it’s less then 3 kmph and my other aunties xtrail it’s 6kmsph again and my partners RCZ 5kmph ah is this so? Are manufacturers doing this on purpose?" - Andrew
Is-the-speedometer-reliable-in-telling-me-my-vehicles-real-time-speed (1).jpg


Yes, Andrew, they are. Doing it on purpose. Because … regulations. 

About 11 years ago the essentially globally homogenised regulations for speedos in new cars changed. Essentially they’re not allowed to under-report your speed. So a speedo cannot display - say - 100 when your actual speed is 105.

This is kind of a good idea because nobody wants to be pinged for speeding on our increasingly over-regulated roads packed to overflowing with over-zealous police officers and private contractor arseholes in camera vans all snapping away, deriving a steady tsunami of government quasi-tax revenue whitewashed with the bullshit premise of safety.

So, on the 1st of July 2006, here in ‘Straya, they updated the speedo compliance regulation called ADR 18. (In the USA it’s likely to be the same regulation called FMVSS-something - the kooky code for Federal Motor Vehicle Safety Standard. In Europe it's called UNECE Regulation No. 39/00. Different names for essentially the same thing. Most of the compliance regs are homogenised worldwide now.)



New ADR 18 says speedos cannot indicate less that the true speed, and over-indication accuracy is limited to a maximum 10 per cent plus 4km/h. This means that at a true speed of 100 kilometres per hour the speedo can’t be displaying 99, or under, but it could be displaying up to 114.

For corroboration, see what the RACQ says about speedo accuracy >>

Also (miracle cure for insomnia alert) the official regulation and specifications >>

Currently, under the regulations, accuracy is tested at 40km/h, 80km/h and 120km/h, and the accuracy of the instruments measuring the true speed during the homologation test must be within half a per cent.

Before 1 July 2006 the speedo accuracy was simply plus or minus 10 per cent - so the true speed could be 100 and the indicated speed could be anything between 90 and 110 - something to bear in mind if you own an older car.

That critical rule re-jig date only applies to our glorious former convict melanoma melting-pot paradise - the changeover might be different in your fine nation, should you be watching from a comparatively cultured, civilised country.

One of the upshots of the latest ADR 18 is that you really cannot blame the manufacturer for committing a speeding offence - because the speedo didn’t make you do it. In fact the speedo is probably slowing you down.

The other upshot is that manufacturers target a suitably high overestimation - in the five to eight kays an hour range - and that allows them some wriggle room on intrinsic speedo accuracy and differences in tyre sizes, etc.

If you've been booked for speeding, there are opportunities to contest it. If you're in this position, check out my interview with expert lawyer Adam Ly on fighting a speeding fine >>



It’s not really a conspiracy to slow us all down - it’s a compliance issue. What galls me is this this: I want to drive on the freeway at the limit. For example, it’s 300 kays door to door from my joint to Canberra - mostly freeway.

If ‘Strayan Prime Minister Malcolm Turnbullshit calls me for an urgent consultation on running the nation, some thorny issue (which tie makes me look like less of a Churchillean git) I have no tolerance for dilly-dallying on these pressing matters of national security.

If I drive at the 110 kilometre per hour limit the drive to save the nation will take me two hours and 44 minutes. And, just like my hero Jack Bauer, I don’t want to keep Bullshit Mountain waiting. The PM needs my help, to prevent Tiegate.  

However, should I happen to be in a car with the least-accurate (but still compliant) speedo, I’ll be staring down at 110 but my actual speed over the ground will be just 96 kays an hour. Transit time: Blown out to three hours and eight minutes. Nobody wants that.

This means I’ll be keeping Bullshit 6 Actual waiting for 24 minutes longer than absolutely necessary. I hate that - also, back in the real world, that’s not driving: it’s just wasting your life bored shitless on the freeway, and I have no wish to do that for a nanosecond longer than needs must.



So, you could use GPS - but not integrated GPS, from the carmaker, because it generally does not display speed. Presumably because manufacturers don’t want to open the floodgate of complaints about speedo inaccuracy from indignant customers, when they see two mutually irreconcilable readings on the same instrument panel.

If you want to drive legally, but at the maximum permitted speed, you can suck a GPS unit to your windscreen. Then, you might compare suck-on receiver’s speed to that of the speedo and derive a correction factor. Or, if you want to use the correct technical jargon, a ‘fudge factor’.

Couple of caveats on this - I’d be doing it on a flat, level section of road, because GPS accuracy is potentially compromised uphill or downhill. The system itself is reasonably robust for the Z-coordinate, but the receiver you use might not be paying that much attention to elevation in practise.

I’d do it at a range of posted speeds, too - because speedo error might not be constant across all operating speeds in your car.

I’d also use a wide-open road without overhanging trees - because canyons and trees that occlude the sky can block the line of sight to multiple satellites and degrade the suck-on unit’s accuracy.

And I’d also be driving at a nice constant speed for several seconds to minimise read errors flowing from the sampling rate of the receiver. Receivers don’t listen continuously. They listen, rest, listen, rest, repeat, for ever. Driving at a constant speed is a safe way to overcome sampling rate issues when there’s a speed change between readings.



If you’ve ever been on the freeway in a 110 zone and blasted over a crest or around a curve and come face-to-arse with a highway patrol interceptor lying in wait, and then - visceral reaction - you look [LOOK DOWN] down there and you’re aghast that the speedo is nudging 120 kays an hour. You’re picturing what happens next...

...and then you might wonder why the blues and twos never actually go on, in this situation, like a deleted scene from Mad Max (you know, the first one, before Mel Gibson emerged like a butterfly from the anti-semitic nutbag cocoon) it’s probably because your actual speed could have been as low as 105 on a perfectly legal speedo. And you thought you got away with it.

You've just been saved by your speedo (and the accuracy regulations).