Photo Credits

Climbers Start Up the Spearhead

Young woman belaying her partner low on the North Ridge of the Spearhead (III 5.6) in Rocky Mountain National Park, Colorado.

© Forest Woodward :

Should You Retire a Dropped Carabiner?

Engineering students play Mythbusters with dropped carabiners

I have often heard other climbers and even teachers at outdoor programs say something like this:

Dropping a carabiner onto a hard surface can cause it to “microfracture” - weakening it so much as to make it unsafe. If you drop a carabiner, even from your waist and only once, you should retire it.

Microfracture is indeed a real thing, but you might need a PhD to understand it. A search for metal microfracture on Google returns scientific journal papers on metals and composite materials, as well as a SuperTopo thread with a debate on whether or not microfracture is real. (Have trouble sleeping at night? Simulating Micro-Fracture in Metal-Matrix Composites, a NASA technical report, should put you to sleep quickly).

A microfracture is a fracture invisible to the naked eye. The fracture could cause a stress concentration and propagate under load, reducing the strength of the material. Metal, or any material for that matter, is imperfect by nature and has defects in its crystal structure, including small cracks [1]. The practical question is not whether an aluminum carabiner microfractures when it sustains an impact, but if that impact actually makes a difference in its strength.

In 2007, when I was a mechanical engineering student at the University of Colorado, friends and fellow climbers Ross Callison, Justin O’Brien, Paul Ginzburg, and I decided to investigate that question. We dropped 30 carabiners from heights of 21, 40, and 109 feet onto concrete, filmed the impact on high-speed video, and tested their ultimate strength. 

The results? There was no difference in breaking strength between brand new carabiners and ones that had been dropped, even from 110 feet. While this should not be treated as a license to recklessly abuse your gear, it seems like you shouldn’t worry about a dropping a carabiner from your waist. Read on to learn more about our testing.

[Editor's note: Strength testing was done using Instron instrument pictured below. See videos of carabiner drops and strength testing from the study at the end of this article.]


Carabiners are made from 7075-T6 Aluminum, which has the following mechanical properties [2]:


  • Elastic Modulus:  71.7 GPa
  • Yield Strength: 462 MPa
  • Ultimate Strength:  524 MPa

Traditionally, carabiners were made by a process called cold forging. In cold forging, a metal is plastically deformed [irreversibly changed in shape by force, Ed.] below its recrystallization temperature [temperature at which internal crystalline structures begins to change form, Ed.] and undergoes strain hardening, making it harder and stronger but reducing ductility [ability to be altered by strain without fracturing, Ed.]. Many modern carabiners are hot forged. In hot forging, a metal is plastically deformed above its recrystallization temperature so that it does not strain harden [3]. Because the ductility of the metal is preserved, hot forging allows for more complex carabiner shapes.

Test Setup

We chose high-rise dorms as our testing site. Our drop heights were determined by who would let us into their room to drop something out of the window – residents of the 3rd, 5th, and 13th stories turned out to be the most amenable. (We did actually get permission from the University to do this, by the way).

Our drop heights were 21, 40, and 109 feet, respectively. The impact surface was 6-inch thick concrete, with a compressive strength of 3500 psi. We tested two types of carabiners: cold forged and hot forged versions of the same model. All carabiners were brand new. We dropped five of each type from each height. We dropped each carabiner only once. We tied a streamer on the end of each carabiner to control the location of the impact – it always landed on the rounded end on the side where the gate opens. We tested 40 total carabiners – five hot forged and cold forged from each height, plus control samples that had not been dropped. We filmed the impact with an Olympus iSpeed High Speed Video Camera at 1500 frames per second.

The carabiners were pulled to failure in an Instron tensile test machine. We followed the UIAA 121/EN 12275 carabiner testing standards and pulled them apart with 12 mm diameter pins [4]. We used a load rate of 50 mm/minute. The upper Instron fixture was always started at the same height. The carabiner was always loaded upright with the gate facing left.

Test site:



Impact Surface:


Impact Surface

Streamer with impact location:



Instron Setup:



Results and Discussion

The average breaking strength plus/minus one standard deviation is shown in the table below. There was no meaningful difference between the control sample and the ones that were dropped.



The breaking strength data came from the force-elongation curve recorded in LabView and exported to Excel. Two sample force-elongation curves are shown below. You can see where the gate engages with the carabiner hook, where the linear elastic region ends and plastic deformation begins, and where failure occurs. Plastic deformation is permanent. If the load is released before plastic deformation occurs, the carabiner will return to its original length.





Thirty-seven of forty samples failed at the hook and three failed at the pin. The failure location is a perfect example of a concomitant experimental variable—one that is measured but uncontrolled. Since one of the control samples failed at the pin and the data do not show that breaking strength was affected by drop height, it seems that the failure location is a concomitant variable and not an indication that the impact weakened the carabiner. The manufacturer confirmed that this variation is normal even in testing brand new carabiners and is not associated with the impact location.

Failure at hook:



Failure at pin:



After being dropped, the spring at the base of the gate on some of the carabiners popped out of place, making the gate non-functional. Specifically, this happened on three of the five hot forged samples dropped from 40 feet, and on four of the five hot forged samples dropped from 109 feet. This was not observed on any of the cold forged carabiners. We were able to push the gate closed on these carabiners before testing them.

High-Speed Film Observations

Due to the camera only being able to record in a very small frame window, not every drop was caught on camera. We saved 15 clips and, from some of them, estimated the carabiner’s velocity just before impact by playing the video back frame by frame. The carabiner did not fall in exactly the same spot each time, meaning it was to hard compare it to the ruler in the background. Additionally, there was some wind at the testing site, so these are very rough estimates. The speeds are shown in the table below. 



The theoretical impact velocity (assuming no air resistance), can be calculated from this equation:

v= 2gh,

where v is impact velocity, g = acceleration due to gravity, and h = drop height [5]. The carabiners should fall a little slower with air resistance, which matches up with the drops from 40 and 109 feet. There is clearly some error in our velocity estimates from 21 feet.

Filming the drops revealed a fascinating insight: the carabiner gate opened on impact. The strength of carabiners with the gate open ranges from 7 - 10 kN – forces that you could potentially generate in a severe lead fall. While most lead falls are in the 2 - 5 kN range [6], that doesn’t leave much margin for safety. Obviously, it’s best if your carabiner gate stays closed. We did not film any wire gate carabiners, but presumably the lighter gate is less likely to open in a fall. It would be interesting to film it and see if that is in fact true. It is scary to think about taking a lead fall and having a carabiner on your protection or rope impact the rock, causing the gate to open. Searching the "Accidents in North American Mountaineering" journals brings up a handful of open-gate failure cases [7, 8], but it’s hard to determine if they are due to impact alone or to the rock wedging the gate open.

Limitations and Conclusion

Any test has limitations, and it is important to be aware of them. First, our data says nothing about repeated impacts. We dropped each carabiner only once. In a very unscientific test in the lab, we dropped various heavy objects onto one carabiner until it was visibly deformed. It broke at about 11 kN, which is too weak for climbing use. This shows that you can actually damage carabiners. Never climb on a visibly deformed carabiner!

We did not test heights above 109 feet. The results of this test cannot necessarily be extended to locking, wire gate, or steel carabiners, as differences in geometry and material could produce different results. We also did not test other metal equipment like cams and stoppers. Some rock types, such as granite, have a higher compressive strength than the 3500 psi concrete that we used, which could result in greater impact forces.

Though a sample size of five is better than a sample size of one, a larger sample would provide more certainty that the results were not random. Our hot forged sample data closely matched the manufacturer’s average, but our cold forged sample data was about 3.5 kN higher than their average. This was possibly due to a calibration error with the tensile testing machine when we were loading the samples. However, the results still seem relevant – one drop did not change the ultimate strength of the carabiners.

Climbing is inherently dangerous, and there is no reason to add more risk by using unsafe equipment. Based on our testing, it seems like you shouldn’t worry about a dropping a carabiner from your waist, as I was once told. However, if you are ever in doubt about the integrity of a carabiner, and especially if there is visible damage, just retire it.

Impact Videos

Calibratation Drop:



Carabiner Drop 7:


Carabiner Drop 8:


Carabiner Drop 11:


Carabiner Drop 12:


Carabiner Drop 13:


Testing Videos

Failure 2 - Hook


Failure 4 - Pin:


Failure 7 - Hook:


Failure 11 - Hook:


Failure 16 - Hook:



  1. Nondestructive Testing (NDT) Education Resource Center, 2001-2014. Iowa State University. 
  2. Forging.  Engineers Edge.  
  3. 7075-T6 Aluminum.  MatWeb.  
  4. UIAA Carabiner Testing Standards.  
  5. Object Falling from Rest.  HyperPhysics.  
  6. Black Diamond QC Lab December 2, 2005
  7. Accidents in North American Mountaineering, Volume 7, Issue 4.  Page 33, 1999.  Fall on Rock, Equipment Failure—Carabiner Broke, California, Lover's Leap. 
  8. Accidents in North American Mountaineering, Volume 8, Issue 3.  Page 95, 2003.  Fall on Rock, Protection Pulled, Carabiner Broke, Exceeding Abilities, Washington, Frenchman's Coulee, Air Guitar.



Jim Margolis

Jim Margolis has worked as a field instructor at NOLS since 2010 in the rock climbing, mountaineering, winter, and backpacking programs. Prior to NOLS, he was an Outward Bound instructor in Montana and Colorado. He graduated from the University of Colorado with a masters in mechanical engineering in May 2008. He is known for eating ridiculous amounts of food and imitating comedian Mitch Hedberg.


The opinions of contributing writers may not be the the same as the opinions of OSI.

Posted by

Jim Margolis

on 1/29/15
Field SafetyTools & ToysResource


I was searching for data on grades of aluminum which suffer from brittle failure at low (sub zero) temps, experienced during mountaineering. That is the temperature region in which cold short occurs. Your study, while well staged, omits the scenario as suggested. Nil response required, thanks.

By Geoff Crumblin on 02/05/2015 at 06:17 AM

Nice writeup. In your introduction, you might want to note that after forming, carabiners are heat-treated, which removes any residual stress from the forming process, therefore hot and cold-forged items are close to equivalent.


By Tom Jones on 02/05/2015 at 08:35 AM

Thanks for pointing that out. 

For those seeking more information, here is an interesting Rock and Ice article that goes into more detail:

By Jim Margolis on 02/05/2015 at 02:37 PM

Hate to be a nerd, but only one of their tests was actually statistically significant. They don’t report their statistics, and that might be why, but if you do them (I did) only the cold-forged group of biners would be considered statistically significant (p=.0375), and then only by very liberal scientific standards. The hot forged is not statistically different from chance (p=0.076) and no scientist would ever base any conclusion upon that. They only test 5 biners in each category….but _even Russian Roulette pans out ok on average 5 times out of 6_. This is not the approach you want to use in testing this type of thing at all. You would want to test large amounts, at a variety of heights, and look for percentages of catastrophic failures, rather than look at a handful of biners dropped from a few balconies. No offence to the study or the authors - I’m sure they are very aware of the limitations of the study. And I think this is all very interesting, but I would not base any climbing, guiding, work or insurance related decisions on this.

By Michael Melnychuk on 02/05/2015 at 04:25 PM

Hi Michael,
It seems like you actually like being a nerd, and I appreciate you lending a critical eye to the article!  I did not do statistics for this article because I thought it would be silly to do them for a small sample size.  We were limited by time and materials in the testing and of course it would be better to test more samples.  In the Limitations and Conclusions section, I clearly acknowledge sample size limitations: “Though a sample size of five is better than a sample size of one, a larger sample would provide more certainty that the results were not random.”  This is best considered an exploratory study, something of sort you might find on Black Diamond’s QC Lab website, not a large study that you could report in a peer-reviewed journal.  Hopefully this comment makes that clear, if it wasn’t clear in the article.

By Jim Margolis on 02/09/2015 at 08:07 PM

Great info, even within the limitations of the testing.  More from On Rope 1 website under FAQ, Myth Busters (#1): “In a test by Steve Nagode, an engineer at the REI quality assurance laboratory, 30 carabiner bodies (half ovals, half D’s) were each dropped six times onto a concrete floor from a height of 33 feet.  Following the drops, their open-gate strength was measured and compared to 30 control samples from the same production batch and which had not been dropped.  The statistical result showed “no loss of strength.”  Inspect any piece of dropped equipment carefully, checking for proper function. Cast metal products are most vulnerable to damage, fractures and cracks. To my personal knowledge, this happened once to a gray cast metal Jumar ascender in the 1970’s. To my extensive knowledge: Drop forged carabiners (and similar gear) have not exhibited this problem.”  About the “fascinating insight,” I thought gates are well-known to open on impact, or when wedged open by a rock, or from gate flutter as a running rope induces vibration. Just smacking the spine of the carabiner on the palm of your hand will make a “click” sound indicating the gate opens and clicks shut just from that small impact.  Still, this insight and the videos are very good to reinforce this risk for all levels of climbers.  Thanks!

By Larry Borshard on 02/12/2015 at 09:50 AM

Thanks. You’re right I actually do enjoy being a nerd. smile Don’t get me wrong. This is more information than we had before and I think that’s terrific, but I work a lot with statistics (neuroscience phd) and I know how unreliable small sample sizes can be. I climb as well and I would not hesitate much to re-use a biner that I’d dropped from 30-40 feet…more so for a cam as they cost a lot…

The reason I came to this page and decided to comment is that your study was shared on Facebook by a professional (IRATA) certified rope access company in Canada, as a piece of evidence for not retiring gear that had fallen a good distance. I mentioned this to them directly as well that making a public statement such as this could result in a huge legal headache for a company in the face of an accident. Obviously you have no control over who cites your research, but it might be good to consider liability issues of people who read it and to extend a firmer word of caution to them. Many of them may not have the training to be able to interpret results such as these.

Really appreciate you taking the time to perform the study and thanks for sharing it too. Increasing our body of knowledge can only help make climbing a safer sport.


By Michael Melnychuk on 02/12/2015 at 01:15 PM

Well, Mr. Nerd, I’m afraid you need to polish your nerdliness.

Jim’s analysis stands as he stated it - there is no meaningful difference observed in this test, between the strengths of the dropped, and not-dropped ‘biners.

Assuming your null hypothesis is that there is no difference, this is _exactly_ what your non-significant P-values mean.  Of course, I’m not sure that was your null, nor am I sure exactly what groups and comparisons you ran, but your interpretation of your statistical analysis appears to leave something to be desired.  It is _always_ valid to say “I can’t detect a difference”  (assuming, of course, that you can’t).  It’s valid to say that, even with an N of 2.  And yes, thousands of scientific conclusions are made daily, on the inability to detect a difference between samples, based on P-values that don’t attain significance.

Does this mean that the presented tests constitute adequate data on which to base one’s personal-protection gear decisions?  That’s up to the individual reader to decide.  I’m just not fond of people attempting statistical-machismo one-upmanship in public forums, especially when their use of statistics is questionable.

By William Ray on 11/23/2016 at 08:31 PM

You are right that n=5 is not big enough sample size to provide statistical significance for such a complex piece.  But you are wrong believing that’s better than n=1.  I hear that kind of comment from biologists a lot, this is the first I got from an engineer… 

But that’s not the biggest problem with your experiment. The key omission is not having inspected it for microcracks between dropping and testing them.  Then you’d have something concrete and quantifiable for all geometries and impact scenarios.

I hate to say that one doesn’t need to test a biner in order to establish the fact that microcracks exist, introduce huge stress concentrations and DO decrease ultimate strength. Myriad of tests on samples of variety of geometries and material processing methods leave no doubt.

Just my 2c after 20+ years of teaching and practicing mechanics of materials in top-tier resarch universities.


By VD on 05/10/2017 at 06:10 PM

@VD—It’s not clear what you’re trying to convey with your assertion that Jim is incorrect with his “N=5 is at least better than N=1” statement.

If you’re trying to say due to the complexity of the object, N=5 doesn’t begin to adequately sample the impact geometry, let alone the distribution of properties of the carabiners, that’s fair, but pretty much just repeating what Jim already said.

I’m concerned however, that to a lay reader, “N=5 is no better than N=1”, implies “N=1 is just as good as N=5”, and _that_ is a dangerously incorrect and inappropriate assertion to put in the minds of lay readers.

If you’d care to provide a bit more clarification there, it might be helpful!

By William Ray on 09/19/2017 at 11:43 AM

Funny, I did almost the exact same test for an engineering project at Dartmouth in 1979.  We dropped 4 carabiners from a two heights, 40’ and 80’ if I recall correctly (the top floor of the science library and Bartlett Tower); two dropped just once and two we dropped repeatedly.  We also found no statistical difference between the dropped and the control carabiners in our small sample set.  I would have liked to see what the results were if the carabiners had reached their terminal velocity, but there was no way to drop them off Cannon cliff without losing them!

By John Middendorf on 09/28/2017 at 02:30 AM

It is sensible to believe that if you drop a carabiner or a cam or critical hardware from a pitch up, the gear could be damaged, and you’d be smart to deep fix design companies

By farak doom on 11/02/2017 at 02:29 AM

Speak Your Mind

Non-member comments moderated before posting. No spam or unrelated links please.

You are not logged in. You can still comment; however you comment will be reviewed by a moderator before posting. Comments must be PG in nature and links unrelated to the discussion will be removed. Thank you for contributing!

Remember my personal information Notify me of follow-up comments?

Please enter the word you see in the image below:

Articles by Category
Free Resources

10 Steps to Better Risk Management - Free DownloadNew! 10 Steps to Better Risk Management - learn quick safety steps for your program. Download free guide. Go»

OSI Newsletter Icon - Signup for OSI's Free Email NewsletterThe Outdoor Safety Newsletter covers risk management for outdoor activities. Go»

Facebook Outdoor SafetyLike OSI on Facebook for current news & discussion on wilderness risk management. Go»

By OSI's Alex Kosseff AMC Guide to Outdoor Leadership, 2nd Ed AMC Guide to Outdoor Leadership
5 Star Rating
Learn More»

© Outdoor Safety Institute LLC • Terms of Use & Privacy PolicyContact Us