Tesla Releases New Q4 Safety Report

JAN 9 2019 BY MARK KANE 42

Tesla cars turn out to have way fewer accidents than the average car in the U.S.

The new quarterly Tesla Vehicle Safety Report brings us some stats about the average distance driven by Tesla cars per accident. According to the manufacturer, the results are way better than average for the U.S., provided by NHTSA.

During the Q4’2018 Tesla registered:

  • Autopilot engaged: one accident for every 2.91 million miles driven
  • Without Autopilot: one accident for every 1.58 million miles driven
  • NHTSA’s most recent data for U.S.: an automobile crash every 436,000 miles

For the Q3’2018 results were:

  • Autopilot engaged: one accident or crash-like event for every 3.34 million miles driven
  • Without Autopilot: one accident or crash-like event for every 1.92 million miles driven
  • NHTSA’s data for U.S.: an automobile crash every 492,000 miles

Comparing Q4 with Q3 (the only two that we have so far – Tesla will provide reports every quarter), it seems that the safety of Tesla decreased. We don’t know the reason.

On the other hand, the average for the U.S. increased. In the case of general U.S. average, Tesla says that NHTSA has released new data for Q4.

Anyway, without Autopilot, Tesla seems to improve safety 3-4 times compared to average.

On paper, Autopilot additionally improves the result, but we assume that most drivers use the Autopilot only on roads which suit Autopilot and higher speeds (higher mileage) like highways, main roads, or in heavy traffic. Excluding other areas (especially city) in Autopilot stats (where it’s not used that often), will obviously show better results.

From a statistical point of view, Tesla stats don’t tell us anything about how much better (or worse in theory) Autopilot is on the particular type of road. We don’t doubt that Autopilot improves safety, but at least this data does not reveal the difference.

General info from Tesla:

“At Tesla, we believe that technology can help improve safety. That’s why Tesla vehicles are engineered to be the safest cars in the world. We believe the unique combination of passive safety, active safety, and automated driver assistance is crucial for keeping not just Tesla drivers and passengers safe, but all drivers on the road. It’s this belief that grounds every decision we make, from the design of our cars, to the software we introduce, to the features we offer every Tesla owner.

While no car prevents all accidents, we work every day to try to make them much less likely to occur. Advanced safety features like Automatic Emergency Braking, Lane Departure Warning and Obstacle Aware Acceleration, come standard on all Tesla vehicles for an added layer of safety beyond the physical structure of each car. Over-the-air software updates allow us to introduce safety enhancements and new features long after a car has been purchased by a customer. And because every Tesla is connected, we’re able to use the more than 10 billion miles of real-world data collected by our global fleet – of which more than 1 billion have been driven with Autopilot engaged – to constantly improve our products.”

Categories: Tesla

Tags: ,

Leave a Reply

42 Comments on "Tesla Releases New Q4 Safety Report"

newest oldest most voted

Great article! The data is young, but this is important data to watch over time. Insurance companies are data-driven. Insurance for my Model 3 LR RWD with FSD is $600 per year. Insurance is also, of course, relative to your coverage, but I was able to match my existing coverage for other vehicles at half the price. The point being, that if this trend continues, the full self-driving option will pay for itself over the life of the car on insurance savings alone. To be fair, Tesla, in general, doesn’t have the “16-year-old statistic” to deal with. That doesn’t answer why there are fewer accidents with Teslas outfitted with FSD. Will be interesting to watch how this changes when the $35k model hits in volume. You would think that the accidents might increase. Then again, it won’t be $35K with FSD so the gap might widen still. Also, FSD gets better with every recorded mile. It’s a brave new world.

It’s not terribly meaningful yet (other than that Teslas are safer than the average car on the road, but that should be expected for newer mostly well-maintained cars), but good on Tesla for releasing this data.

Couldn’t that be simply that AP is mostly engaged in situations when there is overall lower risk of accident per mile?

Agreed, yes, the Tesla cars have fewer accidents with Autopilot (AP) engaged than without, but it is hard to find statistics of rural vs urban accident rates (not fatality rates). You have many more opportunities to have collisions when there are intersections involved.

The data are misleading without further similar data collected for other similar cars driven by similar people.

Tesla’s data tells me the early Q4 roll-out of Version 9 software made, at least MS & MX, cars less safe for manual driving.

It’s not as if we haven’t been saying this:

The company’s aim to prove AP is safer than its drivers makes it less cynical to theorize they would complicate their touchscreen interface. Be objective. Don’t hold ergonomics hostage, to Tesla’s success.

Problem with living in a warm climate…

Q4 brings snow to most of the northern hemisphere where Tesla’s are quite popular.

It was been a warm Q4. Much of the Northeast hasn’t seen its first inch, yet. Maybe that’s also why we buy the cars?

The accident rate increased for all cars in Q4. More accidents happen in winter is a likelier explanation. California’s rainy season (snow season in the mountains) started in November for one thing.

“Tesla’s data tells me the early Q4 roll-out of Version 9 software made, at least MS & MX, cars less safe for manual driving.”

Very strange that you think you can narrow the difference down to just one single variable, and ignore all the others.

This may indicate nothing more than a higher accident rate during late fall/ winter weather conditions while driving. Or it could be the result of Tesla putting the Performance trim level of the Model 3 on sale, since high-performance vehicles attract more reckless drivers.

Or some of both, or some other variable that doesn’t occur to me right off the top of my head.

This data leads to a multivariate analysis that can lead you to incorrect conclusions. For example, AP is used mostly on highways. Highways have a lower probability of accidents. The apparent cause of AP being “safer” than non-AP may be a simple correlation of greater highway use of AP vs city. In fact, I’m certain that this is a significant factor. Same thing with the “Tesla’s catch fire less often than gas cars” statistics. There are other factors, especially the fact that the gas car analysis includes OLD cars, and older cars catch fire more often than newer cars.

I think if Autopilot crashes are one in 2.91 million miles and crashes are generally one in 436,000 miles, then driving a Tesla with autopilot is safer than driving a non Tesla without autopilot. It is false that people only use autopilot on highways, but even if they did, highways are not 6 times safer than other roads. Furthermore, most road fatalities are on highways and the fatality rate of Teslas on autopilot is a heck of a lot lower than the fatality rate of cars on the highway.

I also think it’s clear that Teslas catch on fire less than the average vehicle by a wide margin.

When the data came out in Q3 there was a lot of discussion on this. The data from regional organisations showed that highway accidents were in same order as Teslas claims. The way Tesla portray these numbers are indeed spurious and should be taken with a pinch of salt.

There are generally more deaths on the highway, but usually far more accidents (what’s being measured by this data) in towns (the disparity being that accidents on the highway are normally at higher speeds) for example. kubel is correct in his comment.

That’s not to say AP is not safer, but there are a lot of other factors that are not being considered in these numbers – making them almost worthless for any kind of meaningful dialogue.

I can see clearly that a Tesla is safer than a non-Tesla in this data. I don’t know who specifically is bringing down the nationwide average, but why take the chance? The lesson in all of this is that driving a Tesla is way safer than driving a non-Tesla and autopilot only reinforces the extreme gap between Tesla and other vehicles.

On fatalities, I mentioned those because an accident is one thing, but fatalities are much more serious than say someone dinging you in a parking lot while you’re not even in the car. Tesla seems to have reduced the probability of an accident and also reduced the probability of a fatality during an accident.

The point here is that Tesla’s are safer and claiming otherwise is disingenuous and false. Furthermore, autopilot is objectively safe when used. I don’t plan on driving 2.91 million miles this life time on any road.

What if someone told you that you could drive for millions of miles on a highway and expect to never experience an accident? It wouldn’t make sense to stoke fears or argue about statistical noise in the data about such a system would it, but here we are.

The lesson here is that a Tesla car has been in less accidents than the national average, sure. But the question is WHY? What is the average demographic of the average Tesla driver? Guessing there aren’t many 16 year olds driving round in them for example – as they’re more likely to crash than more experienced drivers then that will push the average down. Do Tesla drivers drive in town more often, are they concentrated in areas with lower crash statistics? etc etc. Those questions are important to understand and analyse the data. As it is the data means nothing. For example, if you or I bought a Tesla today and did the same driving we did yesterday in a different car, would we be less likely to be involved in an accident? The answer is, with that data, we don’t know. The fact is the vast majority of accidents on the road are related to driver error. Just getting into a Tesla is not going to make you a safer driver. Have a look at some actual crash stats, separated by road type. What you’ll notice immediately is that there is a significant disparity between road types. While you… Read more »

In the real world, people don’t ignore the data that is available merely because it’s not from a rigidly controlled scientific experiment, with minimumized intrusion from outside factors. In the real world, people make decisions based on the available data, even while realizing that no study is perfectly free of outside influence.

Claiming that we should ignore the straightforward conclusion of the available data, the data indicating that Tesla cars are actually safer, and assume there is some mysterious and hidden cause… is nothing but a conspiracy theory.

Well, in which case don’t ignore the actual IIHS data then, and drive around in a large ICE SUV instead, it’s safer afterall… They have at least shown their methodology and tried to normalise for external variables.

Not all data is created equal, and raw data (which the Tesla data basically is), is practically useless without further analysis and information on it.

In the real world, people don’t ignore the data that is available merely because it’s not from a rigidly controlled scientific experiment, with minimumized intrusion from outside factors. In the real world, people make decisions based on the available data, even while realizing that no study is perfectly free of outside influence.

Claiming that we should ignore the straightforward conclusion of the available data, the data indicating that Tesla cars are actually safer, and assume there is some mysterious and hidden cause… is nothing but tinfoil-hat nonsense.

“That’s not to say AP is not safer, but there are a lot of other factors that are not being considered in these numbers – making them almost worthless for any kind of meaningful dialogue.”

This is completely false.

A straightforward interpretation of the data shows Tesla cars are not just a bit safer, but much safer, especially when using Autopilot+Autosteer.

Occam’s Razor shaves in the direction of the increase in safety being a result of exactly what it appears to be: The increased safety of using Tesla’s suite of driver-assist features.

An assumption that there is some other mysterious, hidden cause requires one to assume facts not in evidence; to ignore the sound scientific principle of Occam’s Razor. In fact, assuming some other mysterious and hidden reason is wandering into conspiracy theory territory.

Have a look at the links I provided and see why I say that. There is an 8x difference in collision rate between different cars, with a Model S being 3x more likely to have a collision claim than the average car. That’s completely at odds with the “data” provided by Tesla.

Lets put it another way. Why do you think a Tesla is less likely to have a collision/accident than another car? Is it because of the vehicle itself, or is it because of the driver?

Instead of writing lots of fluff claiming I’m wrong, why not actually provide some methodology and evidence to your claims. I’ve happily provided accidents statistics and studies from highway agencies and safety organisations to back up my claims…

“It is false that people only use autopilot on highways, but even if they did, highways are not 6 times safer than other roads.”

Thank you! It gets rather tiresome seeing Tesla bashers try to dismiss the data which shows quite clearly that using Autopilot+Autosteer results in safer driving. The attempt to belittle such statistics doesn’t hold up to scrutiny, or to critical thinking.

This is nothing to do with basking Tesla. The fact you see it that way suggests more about you than the points being made about unprocessed data…

So, Tesla is comparing itself with “average” now? It isn’t an average car by price or its own supporter’s claim.

How about comparing against some of the BEST? Like those models with ZERO fatalities according to IIHS?

In addition, this whole comparison against average is just spin to start with. Tesla is above average in terms of safety features, cost, age and weight. All of those leads to naturally better statistic than “average”. It would be better to compare against similar price, weight, cost and age models and see if it is truly safer.

Of course, that is the beauty of Tesla, they are the BEST marketing spinning company in the world because their marketing spins get picked up by media and spread by its fans for free while an army of them will defend it at all cost.

What are you even saying? Lighter cars get in more accidents? Why? How?

No, he’s saying different cars have different accident rates due to different types of driver and use.

Here’s another example: https://www.cars.com/articles/whats-the-most-crash-prone-car-1420692141385/

The results of that show that something like an Audi RS7 or a Nissan GT-R are almost 8x more likely to be involved in an accident over a normalised distance than an F-150 or a Smart ForTwo. According to Insurance stats a Model S is 3x more likely than the average vehicle to have an insurance damage claim (and the F-150 and Smart are half as likely compared to average). That data is from the IIHS and insurance claims data, with non car related factors accounted for (“operator age, calendar year, density, gender, marital status, model year, risk and state”).

Conversely though a Model S occupant is half as likely to claim for an injury than average, and more on a par with a large SUV than a small car (occupants of lighter cars usually are more likely to be injured in an accident).

Don’t take my word for it, here is IIHS’s own data.


You, also, are confusing injury/fatality rate with accident rate.

The statistics are quite clear that Tesla cars are far less likely than average to get into an accident, as measured by airbag deployment.

You’re also more likely to survive an accident in a Tesla car, but that’s a separate issue. That’s not what’s being claimed here, and conflating the two issues is only causing confusion.

How. I’ve very clearly separated accident rate from injury rate, you know, by using two separate paragraphs.

Keep sticking your fingers in your ears…

“What are you even saying? Lighter cars get in more accidents? Why? ”

Lighter cars tend to have far greater fatalities.

Also, lighter cars tend to correlate to econ boxes. Econ boxes tend to have less safety features that enhance safety, thus more likely to get into accidents if the drivers are the same. Also, younger people or less experienced drivers tend to drive smaller/lighter vehicles which also correlates to higher accident rates.

No, lighter cars don’t cause accidents, but they correlate to higher accident rate due to the difference in safety features and its driver demographics.

That difference in demographics is also applicable for Tesla which has more experienced drivers and more safety features. So, compare a particular segment class with an overall average is an unfair comparison.

That makes no sense. You’re basically arguing that it’s unfair to compare Tesla’s accident rate to other vehicles because Tesla have safety features other vehicles do not. No, that’s the point. Teslas are safer because of their safety features. That’s the argument Tesla is demonstrating.

The demographic nonsense bit is silly too. The number of smaller vehicles that are sold in the US are a tiny sliver of a market dominated by trucks, SUVs, and min-sized sedans, less than 3%. So here you are dubiously arguing about 3% of the market drawing down the average when Tesla is a full 6X safer than average.

But, anyway, if you can find a vehicle with a lower accident rate, you would have made your point. Which vehicles have a lower accident rate?

You are confusing the accident rate with the fatality rate. Just because a car has a higher crash rating, that doesn’t mean it’s less likely to get into an accident; it merely means you are more likely to survive an accident if you have one.

No, the fact that Tesla cars have a very significantly lower accident rate can logically be attributed only to the use of Autopilot and/or Autopilot +Autosteer.

Any other conclusion requires you to assume facts not in evidence, and is little if any better than tinfoil-hat “reasoning”.

What I find interesting is that Tesla did not show mortality rates. It is a a LOT less in Tesla than in other vehicles.

How exactly do you know that the mortality rate is “a LOT less” in a Tesla than in other vehicles when Tesla “did not show mortality rates.” 🤷🏻‍♂️

In the real world, a significantly lower accident rate coupled with superior crash test safety ratings, results in a significantly lower fatality rate.

But again, what is being measured in the subject of the article is the accident rate, not the injury/fatality rate. Some Tesla bashers here are trying to confuse the issue as part of their attempt to belittle the very strong evidence that Tesla cars really are safer, especially when using Autopilot +Autosteer.

They are the safest for the occupant in their class – that’s based on insurance data (IIHS).

Again, not facts.

According to IIHS studies (two separate ones), there are plenty of cars with far lower Fatality rate, ZERO is actually the number. Over 9 models with ZERO fatality. Tesla model S/X/3 isn’t one of them.

Are any of those 10+ models sold in numbers equal to the Model S or the Model X?

If those 10+ models of cars have zero fatalities only as a result of very low numbers on the road, then it doesn’t mean much.

It’s a rate, so number of sales doesn’t matter.

Per the IIHS:
“Rates are given as the number of driver deaths per million registered vehicle years. (A registered vehicle year is one vehicle registered for one year.)”


Technically, it’s the driver fatality rate that is ZERO for those cars in the IIHS studied. IIRC, the IIHS studies reported only driver fatalities. Since some cars carry more passengers than other (ie: minivan vs. two-seat sports car) the only way to fairly compare their relative safety was to count driver fatalities only, otherwise high-occupancy vehicles would naturally skew to having more fatalities than low-occupancy vehicles.

“Anyway, without Autopilot, Tesla seems to improve safety 3-4 times compared to average.”

Most of difference is probably due to Automatic Emergency Braking (AEB) being a standard feature on most all of the Tesla fleet, while the “average” includes a vast majority of cars without AEB.

I’d like to see the accident stats for Teslas with AEB compared to non-Teslas of the same age with AEB.

Sure, but you can’t separate AEB (Automatic Emergency Braking) from Autopilot. Tesla may sell the Model 3 with AEB but without “Autopilot”, yet on the Model S/X, AEB is considered part of the Autopilot suite of driver assistance safety features.

I thought AEB was standard and not part of the optional Auto Pilot suite.

Forgetting all the variables related to the average, what I take from this is AP improved on the accident rate compared to without AP.
So if you can afford it, then buying AP is a good choice to reduce the risk of accident in a Tesla.
Honestly, I don’t think I ever bought a mainstream car and worried about the accident rate. I know how I drive and what my history is. I’m certainly going to be looking for features such as ABS, AEB, etc and so long as the vehicle has those features then the next buying decision is aesthetics and personal taste, not how many accidents has that model has compared to the other models I look at.
I’ll buy the Tesla because my criteria is now EV and Tesla has the best EV available in my country, in my opinion, nothing to do with how many accidents it has had vs other EV’s.