Reasons for accident rate

For what it’s worth: My philosophy was always to push the envelope one item at a time. Eg, if I wanted to increase cross wind landing experience, control all the other variables, so that you are dealing ONLY with increased crosswinds. Don’t attempt tougher crosswinds when visibility is also reduced, or there is a sunglare. Don’t attempt to lower your IFR minimums the first time at night, etc. I found that if I strictly control all parameters of a flight, except the one I intentionally plan to push, that I never seriously compromised safety.

In reply to:

Actually, I was mostly referring to the CAPS. When I first heard about it, my initial thought was “I wonder if that’s going to do more harm than good.” I fly a 182 right now, and one of the negatives of flying a small airplane (among many positive) is the knowledge that there really isn’t as much as we’d like to think between us and a horrific fall to earth. However, it’s probably my constant small fear that keeps me safe to a certain extent. However, the CIrrus seems very attractive to me for a very dangerous reason: it seems to always give you an “out” and I know I’d enjoy flying more if I felt more secure about flying. I’m sure I’m not alone in wanting a Cirrus not so much for its performance, but because I’m essentially afraid of flying and I’d be less so with a Cirrus.

So, the CAPS probably has two effects that mitigate it’s safety benefits:

  1. The Cirrus might appeal specifically to people who inherently feel insecure about their flying ability and perhaps rightly so (low time piilots, inactive pilots, etc.) On paper a Piper cub should be a death trap, but it’s not primary because of the people flying it. The opposite might become true of a Cirrus.

  2. The CAPS (and other equipment) might take away some of the natural fear of flying a small airplane in disproportion to the extant to which it actually takes away the risk. For example, I would never take my 182 into the mountains at night, but I might consider it with a CIrrus. Also, I happen to believe that most GA airplanes are miserable machines to use for serious IFR (one vacuum, lousy autopilots, etc) and so I don’t do serious IFR flying. I might if I felt I could just pull the lever and save myself from any bad situtation I get into.

It’s not that I really think the CAPS is going to save my ass in a redline spiral dive, but it gives you the ability to comfort yourself with the idea that it’s there, and that may present a danger people aren’t totally willing to admit.

Well… I see your point, but… Certainly, there’s no data that I’m aware of to support the fact that we are seeing accidents because people are taking their Cirrus airplanes into situations that they wouldn’t take other airplanes. Of course, it’s possible that some people are flying Cirrus aircraft when they might otherwise not be flying at all… but again, I’m not aware of any like that.
For me, the CAPS was not a big factor in the decision to buy a Cirrus. My decision had more to do with the emotional side (the beauty of the airplane, the clean, roomy and modern interior, the intellectual satisfaction of having a state-of-the-art machine, the cost-benefit equation, etc… however, the parachute IS important to my family and friends, and therefore indirectly to me. I said it wasn’t a big factor, but it WAS a factor; and certainly, there are places I fly in that I’m more comfortable now that I have the CAPS than before… but I can’t think of a single place or situation I’ve put myself in now that I didn’t in my previous airplane, a 182RG; I’m just more comfortable now. A lot of that comfort has to do with having a new airplane, with everything working, with redundant systems, one that I know really well, and many other factors that aren’t CAPS. But that’s just me – there may well be pilots who really are going where they might not have gone before.
Is that bad? Again, if they’re comfortable, capable and current, I can’t see anything wrong with it - in fact, maybe on the contrary… they’re able to expand their experience horizons and be better and safer pilots as a result. I agree with you - most GA airplanes (not all) are miserable IFR platforms, for all the reasons you quoted, and more.
You said, “However, the CIrrus seems very attractive to me for a very dangerous reason: it seems to always give you an “out” and I know I’d enjoy flying more if I felt more secure about flying. I’m sure I’m not alone in wanting a Cirrus not so much for its performance, but because I’m essentially afraid of flying and I’d be less so with a Cirrus.” - Well, if you enjoyed it more, wouldn’t you be able to relax a bit, and be SAFER for it? Obviously, this can go both ways – one could relax and be complacent, or relax and have more time to monitor the systems, the situation, the traffic, etc… but that’s a CHOICE. I don’t know you (I don’t even know who you are), but from your post, I’d guess that you’re more inclined to use the extra safety that’s designed into the airplane to supplement your own safety philosophy, as opposed to using it as an excuse to support being lazy in the cockpit.

I’m sure that in the fullness of time, we’ll find that we have some pilots who get themselves into situations they shouldn’t have come near, for the reasons you outline in your post; and that many others will fly safely in conditions they would quite appropriately have avoided in legacy GA machines.

I’ve never understood people who say, “I don’t want an IFR rating because then I’d fly in bad weather”, or ,“I don’t want a credit card because then I’d spend beyond my means”, or “I don’t want to have the mammogram in case it shows anything”, etc… to say things like that is to cede your own will, your desire and your head to mere circumstance. Surely, if we have what it takes to “earn the ticket” in the first place, we are made of better stuff.

Once again, sorry to ramble… and still just MHO.



I see your point. Assuming that there really is something about Cirri or their pilots (and that we just haven’t had a 1 in 10 chance string of bad luck with accidents) then I don’t see how you can avoid making one of two possible conclusions:

  1. There is something about the airplanes that makes them hard to land.

  2. There is something about people who buy Cirrus aircraft that makes them think (incorrectly) they know how to land.

I’m hoping it’s the second, because that’s probably easier to fix.

Finally, to close out the whole statistical significant debate: nobody was suggesting anyone here did a fully controlled statistically analysis. However, that doesn’t proclude making rough conclusions. One such valid conclusion is that these first generation Cirri are not going to go down in history as the airplanes which solved the GA accident rate problem.

Given my high hopes for the Cirrus, I was expecting to see us with a REAL statistical significance problem in that there would be NO accidents! Given that we’re obviously not at that extreme (or anywhere near it) I thought it would be nice to talk about why, since I think undeniably statistically INsignificant numbers of accidents are what we’d all like to see.


"Regarding overconfidence in the advanced avionics: most newly produced airplanes (new designs or not) have highly capable avionics/autopilot packages comparable to those offered in the SR2X planes. In addition a growing number of planes in the used fleet have been upgraded with things like GNS430’s, STEC autopilots, even Sandels etc. I do not myself see, nor have I read any indication, that this more modern equipment increases the risk of mishap in the GA fleet as a whole. For any pilot who flies the equipment often enough to maintain reasonable familiarity/proficiency, I think it can only help. During the first couple dozen hours it may present a learning curve to be climbed (mixed metaphor?), but after that I think not.

Speaking as objectively as I can, I agree with Clyde that the number of SR2X in the field and hours flown is far too small to draw any meaningful conclusions.

On the other hand my own personal opinion–subjective and quite possibly wrong–is that in comparison to the standard Cessna/Piper issue the Cirrus planes present a sufficiently different (not inferior–I love landing the SR20–but just different) landing experience to transitioning novice and/or low time-in-type pilots that over time we may see a small but significant increase in landing mishaps compared to those other planes. These would be generally minor things: tail strikes, runway overruns, porpoising/prop strikes, etc.: dinged metal/fiberglass that cost $ to repair but few or no injuries. "

I absolutely agree with Clyde in that the numbers are too small to be meaningfull As for the no evidence that advanced avionics does not lead to accidents, I totally disagree. There are legions of stories of runaway autopilots causing crashes or in flight breakups. I believe the Malibu had an issue there. How about pilots flying ‘gps direct’ VFR through terrain, etc. This stuff DOES happen. Statistics, I don’t know, but if it happened once, then it does happen.

More important questions are; “Do the advanced avionics prevent more avionics than they are factors in causing?” and, “How can we prevent the advanced avionics from becoming a factor in (any) accidents?”

My point is NOT that the advanced avionics are in and of themselves dangerous, but that it is the pilot that makes the flight safe. The avionics are tools that can be used or misused. They should not a crutch for lack of skills, currency or capability. In the right hands (and minds) they are truly wonderful and a big safety enhancement.

As for the landing experience. Yes the landing experience of a Cirrus is different from a Cessna, Piper or whatever, but not better or worse. I agree that most of the transitional type loss of control upon landing accidents should be minor, but not all. Actually, isn’t that exactly what we are seeing?


I think the ‘totalled’ aircraft was an SR22. The SR20, that I think you are speaking of, was a plane that had a forced landing and suffered very repairable damage to the wings and cowling.

As far as finding the data for actual Cirrus accidents, it is tough to nail it down. I’ve learned of a couple of Cirrus incidents in the FAA Preliminary Accident Reports, which only stay posted for 10 days or so. Then, searching the NTSB permanent reports, I’ve found some Cirrus incidents…only to not be able to find them on subsequent searches (I use Cirrus Design, SR20, and various other combination searches and get different results each time!). Also, searching the historical pages on this web bulletin board will reveal discussions of other cirrus incidents, such as the numerous botched landings, and last summer’s weather related fatal accident out west (and the SR-22 weather related nonfatal accident down in Georgia). I suppose some botched landing incidents never make it into NTSB. There is also another unofficial SR20 website at, which is where I learned of a total engine failure that occurred last November. I corresponded with the owner of that one and he said there is no final report out yet from the NTSB, and the cause is unknown. (I believe it definitely was mechanical…not fuel or weather related).

In the end, unfortunately, because of the inadequate NTSB databases and search reliability, much of my knowledge of Cirrus incidents is based on memory. I think that is one reason a bulletin board such as this is critical…there’s a lot of knowledge among the various participants, who seem willing to share it.

Clyde -

Actually, the statistical situation is not as hopeless as you indicate. The data is not great, the sample size is relatively small, but the accident rates are signficant in a statsitical sense.

If you use the NTSB database and accident definitions and couple it with CDs estimate of hours flown from the warranty and repair data they get, you can get an estimated NTSB-accident rate for Cirrus pilots and for non-Cirrus GA pilots from the 2000 Nall GA accident report (same database).

On that basis, the last time I did the calculations, Cirrus pilots had an accident rate that was significantly higher than the 7 NTSB accidents per 100khrs flown for the GA fleet - and the difference was statistically significant at the 90% level. Not a sure thing, but enough to worry you.

I also tried to correct for time-in-type, the most obvious difference between Cirrus pilots and non-Cirrus pilots. While GA and military statistics indicate that the first 200 hrs in type have a pilot accident rate as much as 5 times time the “over 200 hrs in type” rate, there isn’t sufficient data on either Cirrus or GA pilots individual time-in-type to put the numbers on a consistent basis. So, it sounds as though time in type is part of the difference, but its just speculation.

However, if the issue is time-in-type, the accident rate for Cirrus pilots in total should decline significantly next year as a great percentage of the Cirrus pilots are no longer new in their aircraft.

You’ll also notice that the above discussion is phrased in terms of Cirrus pilots. With 75%+ of GA accidents being attributed primarily to pilot error, it is primarily the pilot we are discussing here - not the aircraft.


Pete, you make some well-thought out comments. I disagree, though, that the SR2x aircraft demand more of their pilots than, e.g. a 172. I find the SR20 much easier to fly than the assortment of other single-engine planes I’ve experienced. Landings are just different to some, not more difficult, and en-route the avionics make flying a breeze.

I sort agree with you Pete. I think the older we get (over 60) the slower we should go. Things happen a lot faster sometimes…than the old brains can think. It is just part of life. Once I get uncomfortable with my thought process, I’ll trade my Cirrus for an AirCam. I doubt I could hurt anyone but myself at 40 MPH. Having said that, it seems that the few Cirrus accidents have happened because of bad judgement on someoneÂ’s part. Bad judgement comes from inexperience and/or lack of understanding of the situation. I think half of the Cirrus accidents were the results of things beyond the control of the pilot, but they clearly count against us with the insurance companies.

Bob, could you share the actual numbers that you used?

Incidentally, I agree with you that we might expect to see a higher rate of non-fatal accidents (especially landing accidents) during the phase when the majority of Cirrus pilots have low time-in-type. This type of accident is actually not too worrying - it’s the fatal accidents that really are a concern.
I’d like to see the data that you used to arrive at your conclusions, though.

Bob: According to AOPA’s ASF data the accident rate for complex aircraft with pilots’ experience at the 200 hours time-in-type appears to be about double the ‘stabilized’ accident rate. If you go to 100 hour or less time-in-type, the numbers seem to reflect that the accident rate may be as much as 8 or more times as high. (Clyde - These accidents are characterized as ‘serious’ so I do not think it holds that these are relatively minor landing scrapes.)

Now if my memory serves me correctly, always a risky proposition, the venerable 172 has a accident rate of between 4.5 - 5.0 per 100,000 estimated flight hours. Now for Cirruses we have to make some assumptions. Cirrus has delivered just over 300 aircraft. Since the average GA pilot flies less than 100 hours a year we could assume that nearly all Cirrus pilots have less than 200 hours time-in-type. However, you and I both know that the Cirrus drivers seem to really love their planes and fly the cr** out of them. So lets just for arguments sake say that 45% have 100 (or less hours), 45% have 200 (or less hours) and the remaining 10% have more (let’s say 300).

Using this total SWAG method, the Cirrus fleet would have about 50,000 hours. (Personally, I think the aircraft have that many or a few more hours, but the pilots would have less allowing for multiple pilots per aircraft.) Therefore, the average Cirrus pilot would have about 165 hours. (Again, I think that number is high).

Using your data of 7 accidents in 50,000 hours, the AOPA’s ASF data, and a comparison to the 172 using 5 accidents per 100,000 hour stabilized, we can make the following conclusions: Adjusting the C-172 for 165 hours time-in-type, we would expect an accident rate about 4 times higher than the stabilized rate or about 20 accidents per 100,000 hours. The Cirrus fleet rate is only 14 per 100,000 hours.

Now, the last think I want to do is say these conclusions are facts, but on the face of it, using reasonable estimates, the Cirrus fleet rate may actually be well below the expected curve after adjusting for time-in-type. (Yes, I know I made a few totally inappropriate leaps here such as citing complex aircraft rates and then applying them to the C-172, but that is not the issue. I think that if we compared the Cirrus fleet to the Mooney or Bonanza fleets, the numbers would be even more convincing. I just don’t have them.)

I absolutely agree with Clyde that we need more time and data points. I also agree with you that the current data may be statistically significant, but I do not believe that your analysis controls for time-in-type.

So what is my conclusion? It is not the rate of accidents that is important, nor the statistics. Every accident involves at least one person. We need to have less people involved in accidents. If this discussion helps in that area, then it is an extremely important and worthwhile endeavor. Let’s not get sidetracked on the statistics, but concentrate on reducing (or eliminating) accidents.


Post deleted by Bill_Dobson

Clyde, would you care to expand on the statistical analsyis that you supposedly did. What method was used? Not being a statistician, but having studied statistics to a certain degree, I am interested in your methods. Certainly, with regard to fatal accidents a N=1 is very difficult to use by almost any statistical method to establish scientific significance. Unfortunately, it will not be until N equals a greater number, that we will be able to achieve a more reliable level of statistical significance. In addition, if N stays at 1 for quite some time, we will have to accumulate several hundred thousand hours before statistics can be compared.
Certainly, there are some very impressive statistics available that show a steep drop off in accidents as pilots accumulate 200 hours in type. With this in mind we may expect a higher accident rate for Cirrus for several years and not just one or two. Remember production is still ramping up and there will be many new Cirrus pilots flying soon, including myself (I just started last July and have 95 hours and am working on instrument).

Marty -

As I said, it appears that time-in-type is (part of) the problem.

While your data is plausible, it’s no more than a guess. Which is about as far as I got when I looked at it. The biggest problem with the data is that the time-in-type numbers aren’t accident rates (accidents/100khrs). Instead, the data I’ve seen are % of accidents involving pilots with <100hrs etc. Unless you know the number of hrs flown in each category,you REALLY have an apple and an orange. The biggest unknown being how many pilots stop flying in each category.

So, I agree, it appears that ime-in-type is part of the problem. And, if we’re right, the reported accident rate should come down progressively as pilots become more experienced.

But, the bottom line for our pilots is the same - flying is a potentially dangerous activity and, for now at least, its about twice as dangerous for us as for the GA population in general. So, if ever there was a time to be a careful pilot, now is it!

Post deleted by Bill_Dobson

IMHO Flying full steam into a mountain because you aren’t instrument rated yet you are flying in the soup IS bad judgement!

How about getting so far behind the airplane because you are scud running and forgetting about fuel management?

How about taking an airplane home as the PIC when you aren’t confident enough to land the plane? Why couldn’t you say… “I would like a little more training please, because I am not comfortable yet.”

These are my take on the reasons for three of the accidents from last year.

John “JT” Helms
Branch Manager
NationAir Insurance Agency

I agree with Mr. Helms TOTALLY. All the talk about a so- called increased accident rate leaves the impression that there is something inherently wrong with the SR 2X airplanes. I, quite frankly, do NOT think the SR 20 is more difficult to fly than a Cessna 172. It STILL has fixed gear, it STILL has one power lever and it lands and takes off at only slightly higher speeds. Sure it cruises faster but with the fixed gear it slows down easily and speed management does not seem to be a big issue. I also think it lands EASIER than a 172 and certainly is better in crosswind operations. The avionics are dazzling and one could get caught up diverting your attention span to the MFD and Garmin displays. But, otherwise, there is little about the plane that is that complex.

None of the accidents to date have shown an inherent airplane cause. All seem to be “pilot error” accidents that could have happened in ANY plane.

WE do all have to look at ourselves as the “weakest link” in the accident chain. The tendancy today is to blame all others except ourselves for everything anymore, particularly when there is the potential for a deep pocket lawsuit. But the reality is that training and careful flight planning were the ONLY way to avoid tne SR2x accidents so far except the cases where the engine was the problem.

I don’t think anybody was blaming the CIrrus. I think we all understand that the problem is the pilots. I was simply trying to explore possible explanations of why there might be a correlation with Cirri and bad piloting.

For the record, I do think we go too far with the notion that the pilot is usually to blame. You can always see something the pilot did wrong, but a good machine is tolerant of the fact that people make mistakes. I’m not suggesting that Cirri are worse in this regard that other airplanes, but I am suggesting (among other things) that Cirrus Design perhaps hasn’t made the strides they think they have in this area.


I’m not sure why you would insist we talk about cofactors, when all people have been talking about is overal accident rate. Nobody is yet trying to make the case that the Cirrus is inherently unsafe. I think we all know it’s not.

As for the significance of the current data, would you be happy with 100,000 flight hours of data? Well, even if there were no accidents during those hours, the existing accidents would already put is about par with the fleet. So while you probably can’t say with significance that the Cirrus has a much higher rate, I think you can certainly say for sure it doesn’t have a much lower one.


Bill -

If you really want an answer to your question (which I doubt),

  • the hypothesis tested was a simple one - what is the probablility that a Poisson process with occurance rate of 7 per 100khrs would generate 4+ accidents in 27khrs.

  • probability ~ 10%.

Rough - but probably enough to make you worry that our pilots are, in fact, experiencing a higher accident rate than the GA average. Which was all that was claimed.

No attempt to assign causality for all the reasons other have stated.