Crash-Test Ratings for Dog Car Restraints, Explained

By Jessica Ennis, DVM, CVA

Travel · August 10, 2026 · 5 min read

Crash tested is a specific, checkable claim, not a marketing word, and most restraints on shelves have never earned it.

A ten-year-old Weimaraner named Otto rode in the cargo area of his owner’s wagon for years, no restraint, just a folded blanket. Otto’s owner brought him into my exam room after a hard highway stop sent him sliding into the back of the driver’s seat. He was bruised, not broken. He was also lucky. That visit is the reason I ask every client with a car-riding dog the same question before we talk about anything else: is the restraint actually crash tested, or does the packaging just say “safety”?

Start here — what to check before you buy:

• Look for Center for Pet Safety certification, not just the word “crash tested” on the box.
• Match the restraint to your dog’s actual weight, tested products are rated to specific ranges.
• Check the tether length on harnesses — testing standards call for no more than 6 inches of slack.
• For crates, confirm anchor strap and door-latch testing, not just cage-bar strength.
• Don’t assume a carrier that fits under an airline seat is crash rated for a car — those are different tests entirely.

Why “crash tested” is a specific claim, not a marketing word

Most dog restraints sold in the US are never independently tested against a real crash pulse. A manufacturer can call a harness “crash tested” if an engineer tugged on it in a lab, and nothing requires them to say what that test actually simulated. The Pet Carrier & Crate Crash Test Research: Center for Pet Safety is the organization that changed this, working with MGA Research Corporation — an independent lab under contract to the National Highway Traffic Safety Administration — to run pet restraints through the same kind of sled testing used for child car seats. That distinction matters because a sled test at a simulated 30 mph impact is a fundamentally different bar than a manufacturer’s marketing claim.

What the actual harness testing found

The 2013 Harness Study Results tested eleven harnesses on crash dummies representing a 25-pound terrier mix, a 45-pound Border Collie, and a 75-pound Golden Retriever. Only one harness in that round, the Sleepypod Clickit Utility, passed without catastrophic failure across all three dog sizes. The rest showed stitching failures, hardware separation, or buckle release under load — meaning the dog would have been thrown from the restraint in a real collision. That’s not a minor manufacturing gap. It’s the difference between a dog staying on the seat and a 45-pound animal becoming a projectile in the cabin.

I’ve watched this play out from the other side of the exam table, too. A client of mine had bought what she believed was a “crash tested” harness for her 60-pound mixed breed, based on the word on the box alone. When I asked her to pull up the actual test report, there wasn’t one, just a product photo and the phrase repeated in three places on the listing. I’m not going to pretend that every owner has time to dig through crash test PDFs before a road trip. But I do tell clients this: if a company can’t point you to an independent test report with a named lab and a named protocol, treat the crash-tested label as unverified.

Crates and carriers get tested differently, and results diverge

Crates and carriers face their own separate testing, and the 2015 Crate Study Results showed the gap is at least as wide as it is for harnesses. Only one crate held up cleanly under a 30 mph test in that round; several others, including products from established outdoor and travel brands, showed anchor pin failure, structural cracking, or the crate striking the seatback hard enough to injure a passenger. If you’re weighing a crate against a harness or a barrier for your setup, the tradeoffs are different enough that they deserve their own comparison — see our guide on dog seat belt vs. crate vs. barrier for how to choose between them.

Where the evidence is strong, and where it’s thin

The evidence is strong that untested restraints fail at meaningful rates under simulated crash conditions; that’s a repeatable, independently observed finding, not a single anecdote. The evidence is thinner on how those lab results translate to real-world crash severity and speed variation, since most owners don’t crash at a controlled 30 mph with a straight-on pulse. It’s also worth naming the counterargument some owners raise: no restraint is required by federal law, and plenty of dogs ride unrestrained for years without incident. That’s true, and it’s also survivorship bias. A restraint’s value shows up in the crash you don’t expect, not the years you didn’t have one.

What I actually tell clients

For a dog with any joint stiffness or mobility limitation, an untested or poorly fitted restraint carries a second risk beyond the crash itself: a hard stop or sudden lateral force can jolt an arthritic hip or a healing joint even without a collision. That’s not a reason to skip restraint. It’s a reason to prioritize fit and tested performance over price. If you’re setting up a car for the first time with a new dog, our guide to traveling with a puppy for the first time covers restraint basics alongside broader trip prep, and if flying is also on the calendar, the carrier requirements differ enough from car crash testing that they’re worth reading separately in our guide to airline-approved dog carriers.

I think back to Otto more than most cases from that stretch of my career. He recovered fully within a couple of weeks — some soreness, a course of anti-inflammatories, nothing worse. His owner switched to a crash-tested harness within days, before I’d even finished explaining why the blanket wasn’t restraint. That’s the response I want from every client: not guilt, just a fast correction once the actual test data is in front of them.

Where I land

Buy from the Center for Pet Safety’s tested-and-passed list when you can, size the restraint to your dog’s actual weight rather than a general range, and treat any “crash tested” claim without a named lab and protocol as unverified. It’s a short checklist. It’s also the difference, in the cases I’ve seen, between a bruised dog and a badly injured one.