I’ve been hesitating to post this, as I’m really not trying to be flame bait, but
I’m concerned and want to get a sense of what others here are thinking. I’m
a position holder on an SR20. One reason I wanted to buy the SR20 was that it
seemed like a relatively safe plane for its level of performance. (I realize that a
parachute is not going to save me in a low altitude spin, but it would be nice to
have in a mid-air, an IMC engine failure, or a structural failure…)

I’m concerned that there seem to have been a fairly high number of Cirrus accidents
and fatal accidents relative to the total number of Cirruses out there. Does it just
feel that way (maybe, because I’m a position holder, I’m just noticing these accidents
more, and incorrectly assuming that the relative number of, say, C172s is so high that
the fatal accidents in these are still much rarer on a percentage basis)?

What are those who have Cirruses and those who are waiting to get one think? Is the
accident rate higher than expected for the number of planes? If so, why?

Again, this is not a troll, and is not intended to be a call for hate mail. I’m a low time
pilot who is wondering whether he should rethink his decision to buy an SR20…

Thanks –


whenever one is drawn into these types of discussions, there is the inevitable statistical compilation of the accident rate of the cirrus vs. other types of crafts. i became a cirrus position holder at a time when there were no statistics, and mostly because it “felt right” at the time. statistics aside, its losing its feel.

in fairness, some of the accidents may have been the result of pilot actions and, perhaps, poor decisions. there are many individuals on this board far more knowledgeable than me, and could better categorize the accident causes. nevertheless, for whatever reason, the actual experience of the plane does not seem to bear out the implied safety at the time of my purchase. am i concerned. you bet.

Flight Safety, a upper scale flight training organization uses the motto “The most important safety feature in an aircraft is a well trained pilot”. They’re right.
I don’t think the Cirrus is more or less safe than any other GA aircraft. Certainly the CAPS system can’t prevent an accident (it may just help you walk away from one). The high tech avionics are very useful if you know how to use them but can be very distracting if you don’t.
Airplanes don’t fly into mountains. Pilots fly them into mountains. Airplanes don’t fly into thunderstorms. Pilots fly them into thunderstorms. Airplanes don’t screw up landings. Pilots do although admittedly some aircraft are easier to land than others.
The point is that if you are a conservative, careful, proficient pilot you and the Cirrus will get along fine. If you think you’ll be safe because you’re flying a Cirrus you may get a rude awakening.

I can’t speculate on the meanings of a % numbers comparison, because, as I’ve stated before (read the thread from just before the kentucky accident), the actual numbers are too low for a valid statistic inference to be drawn (sample size is too small). They do SEEM high because, being a new aircraft type hyped for its safety, any accident / incident gets lots of media play. Also, if you’re getting your impression from this forum, remember that we’re intensely interested in these things, and so tend to dwell on them, which tends to imply more significance relative to general aviation as a whole.

That being said, I find my SR20 to be the most stable, docile, forgiving aircraft I’ve flown. Without hard evidence to the contrary, don’t blame the airframe for the accidents! (now, if I can just stop that first bounce on heavy crosswind landings, I’d be perfect! [H] )

You are right to question buying a Cirrus, but not necc. because it is an unsafe airplane. It may be, or it may be the pilots, many of which are low time pilots who think that the Cirrus will save them from having to excercise judgement and skills.

The people who say the accident rate isn’t “significant” are flat out wrong. Statistical significance refers to the confidence interval of the population (the “truth” about Cirri) vs. the sample (the current data to date). While the small number means that we have a large rangle, unfortunately the fact that Cirri have already crashed so many times means that the confidence interval, though large, covers horrible territory.

As I’ve said before, and ESPECIALLY with the most recent fatality (flat spin out in Syracuse), the Cirrus fleet would have do perform a statistical miracle to even make it to average by the time the total flight hours reaches a large number.

The average number of fatal accidents by instrument rated pilots flying something like a 172 is about 0.5 per 100,000 hours. Cirri have already had 2 fatals in what is probably significantly LESS than 100,000 hours. So the fleet could fly over 300,000 hours in the coming years with absolutely zero fatalities and still only do about average.

Sounds like maybe there’s something worth talking about here. The aircraft that was supposed to reinvent GA safety might just reinvent it, alright, but not in the way they’d hoped.

Trust me, I’m not happy about this. I had my heart set on getting a Cirrus in the not-so-distant future.

No airplane can make you a better or safer pilot–that is up to you, and is a matter more of judgment & discipline than of physical skills. Having experience in a number of GA aircraft including what are in my opinion the two safest (one of these is the Cirrus), I can say that there is not a single one of them that cannot rapidly come to grief because of poor pilot judgment. What does the Cirrus have going for it? Pros: It is less complex to actually fly & operate the plane. Its airframe and crash-protective interior construction is stronger than the average aluminum GA plane. The wing design makes its stall behavior more benign and offers a degree of spin resistance. The moving maps offer all the situational awareness one could ever hope for (although any GA plane can be so equipped). And, of course, CAPS–the small plane version of the ejection seat. Cons: Avionics complexity requires considerable pilot proficiency. The SR20’s performance decays rapidly when you’re high and hot (not such a concern with the '22). Approach and landing speeds are fairly high, similar to high performance, complex singles and some twin trainers. Engine out emergencies are not as forgiving because of high best glide and approach speeds and relatively rapid descent rates, for example 94 KIAS/900 fpm down in the SR20: better decide NOW where you want to put it down and you may need a fair amount of open space to do it. When I weigh these factors for myself, I still conclude that Cirrus makes one of the top 2 or perhaps 3 safest planes–but does not necessarily stand out in this particular elite group. I would be proud to own one but am otherwise “betrothed” right now. I have enjoyed flying the SR20 and can certainly recommend it as a very good choice.

I somewhat agree and somewhat disagree.

I think the Cirrus is a much safer aircraft than others. However, statistics can always be used to paint a different picture. I don’t think there are any statistics that say an SR20 in particular is a dangerous airplane compared to others.

However the Lexington accident and the squawks on this board do indicate there is an instrument failure issue.

Many failures of vaccuum pumps, transponders, autopilots, turn coordinators, etc do SEEM to be more prevalent than other aircraft. Are they in reality? I don’t know. Certainly MY airplane has had a tremendous number of instrument failures (the TC and autopilot and transponder)

Instrument failures are a proven element in other IMC crashes. We cannot assume that our planes are safe until we can make sure our instruments work reliably.

On the flip side of this, the Cirrus undoubtedly has many safety features (assuming the instruments work of course.)

For example, my SR20 B model with a Sandel and Stormscope upgrades provide a greater level of safety to a trained pilot. Ah, but that’s the key. I feel very confident in my abilities to pilot this aircraft and its systems.

For example, dual vacuum pumps. This is excellent. This is a very important safety feature that i don’t think anyone can argue against.

Battery backup TC. Another important safety feature, not present in other similar aircraft.

In my case, a Sandel EHSI. I find this device a fabulous safety feature. Why?

  1. Features. It has the features that I think are important, moving map for situational awareness, stormscope information, RMIs, great HSI presentation, color coding of source input, numerical values for heading, track, waypoint, etc.
  2. Scan. All of those features are right where you need them – in your scan.
  3. Electrically driven. This is a great feature. If I lose electical power, this device goes down, but I still have my TC (with the flip of a switch) and a compass. If the vacuums fail, I still have this instrument and the others.

Dual alternators. Nice to have option for redudancy.

Garmins. The garmins are a safety feature in my opinion, but ONLY if you are good with them. Why? Pull up any approach to the nearest airport, quickly. Shoot an approach more comfortably and easily in turbulent IMC. I could go on. There is NO substitute for lots of practice and pushing yourself to the limit. You need to be able to handle all of the scenarios that I and others have posted on this forum quickly without fumbling in order to have this device be a safety issue.

Flyability. The plane flies very nice with a nice high wing loading keeping it stable. BUT this is also a liability. The speed at which the plane lands especially in an off airport landing, will have considerably more kinetic energy to dissipate than a 172. In that regard, this plane is far more dangerous.

Stall handling. The plane stalls nicely with aileron control. However, that’s probably not our big killer. The stall spin on base to final is and the characteristics of this plane aren’t going to be any more forgiving in this regard compared to another aircraft.

Stormscope. A great safety feature. Or is it? Will people fly where the area forecast has predicted T storms? Yes. Is that OK? Probably, but will it cause more accidents? Unknown.

In commenting on the specific question. Is this plane more dangerous. Only YOU can answer that question. As a personal example, I have no plans to become multi engine rated and purchase a twin. Why? I know that I will not be able to maintain the proficiency required to make me SAFER in a twin than in a single. Statisitcally, you’re more likely to die in a twin from an engine failure than a single. Those that are proficient, have no problem. Those that aren’t will.

I am staying current in the Cirrus, but I also want technology to help me where it can. If Mr. Itzak Jacoby (Bonaza instructor, CFII, ex-military pilot) can die in a plane crash because of spatial disorientation, then certainly I can too.

As such, for my money, the Cirrus is the BEST choice to give me a hand when I need it most. But only if you respect it.


Well said, Jerry. The accident rate in GA is depressing, whether you look at the SR20 or the whole fleet, but year after year, the #1 cause of accidents is pilot error. Safety is up to us.

I flew with a Sandel HSI in my Baron for 135 hours. I bought it because of the safety issue, and I believe it contributes to situational awareness. The day I sold my plane (after moving to high priced California), I got a call from the new purchaser that he could not get the display to come on. Turns out the light was bad. This is well before the reccomended change out period of two hundred hours. Fortunately it was on the ground but it really makes you rethink the “safety issue”.

I am seriously looking at a SR20. Relative safety is number 1 issue. Most of my flying has been in a Katana. I believe it is the only GA fleet with no fatals in the US ( 1 in Canada?). I assume the very low stall speed, high glide ratio, light weight (lower kinetic energy in a crash) and very strong composite structure are the main reasons. Was wondering how much more of a downside the high stall speed is the SR20 relative to all the other safer elements in the SR20 (stronger airframe, seats, more reliable engine, back up systems.)

To take a stab at an answer to my own question.

On a windless sunny day over flat terrain the Katana wins. On cross country trips over unfriendly terrain and in any kind of bad weather the SR20 wins. (The Katana of course can only fly VFR)

In reply to:

The average number of fatal accidents by instrument rated pilots flying something like a 172 is about 0.5 per 100,000 hours

My recollection is that the fatal accident rate for the 172 is about 0.5/100K hours, not just for instrument rated pilots. But the rate for other, high performance, aircraft is much higher. I don’t have fatal accident rates to hand, but the collective TOTAL accident rate for a group including C182, C210, Commander 114/115, PA32, PA24 and M20 is 7.46/100K hours. Without knowing the number of hours flown by the Cirrus fleet, it’s hard to give an exact comparison, but I figure it must be close to 100,000 hours now, and there have been 6 accidents, excluding one on a pre-production aircraft. So that’s not too far removed from the figure for the other types, even before factoring in the collective low time-in-type for Cirrus pilots.

Comparing the fatal accident rate is less meaningful due to the lower number of accidents. Whether it’s “significant” or not depends on your definition of “significant”. From a statistical point of view, the numbers currently available are inadequate to make accurate PREDICTIONS about what the future accident rate of the Cirrus fleet will be, and the only discernable pattern in the accidents is that of pilots with low time-in-type, which has been found in the GA fleet in general to be a good predictor of a higher accident rate. That is a problem which is self-limiting, and the only message for owners is “get plenty of high-quality training, and make conservative decisions”.

I’d like to take exception with the model and math you used. Any parametric multivariate anova requires a minimum of 20 samples per treatment cell, and sometimes a lot more (you have to run a power test before you can do the actual analysis). Non-parametric approximations are more flexible with the sample n, but are much less rigorous procedures. For several reasons, simple sample distribution tests such as you describe are not appropriate to model such a complex issue as Cirrus airframe safety because they introduce untestable main effects assumptions. To avoid being tedious, I’ll simply say that I can easily identify at least 5 variables that would produce main effects and interactions in a decent model. If you wanted to do a more realistic test and still include your assumptions, you’d have to tailor the “GA” population by eliminating all aircraft older than the oldest Cirrus, all pilots with more in-type time and training than Cirrus pilots, and from the remaining population, randomly pick aircraft such that total flight hours does not exceed that of all Cirrus flight hours. Then, compare the accident rate of that population to that of the Cirrus population. This still would not be an accurate comparison, but it would be a lot fairer than the comparison you’ve chosen to make.
Humm… another method would be to generate the population using only data covering the equivalent aircraft over the equivalent time period. (i.e the first 340 Bonanzas, the first 340 Cessna 172s, the first 340 Piper warriors, etc. during the first 4 years from the time the first one of each type came off the assembly line.)
Unrestricted population comparsions are scientifically unsound. Many of my biologist colleagues make the same mistakes, and they end up paying consulting statisticians lots of money to salvage their data by converting bad a priori models into marginal a posteriori models.

Perceived risk and actual risk are NOT equivalent, no matter how badly the fear mongers want them to be.

… the most recent fatality (flat spin out in Syracuse)


Please stop whatever you are doing right now, and contact the NTSB. Since you seem to know exactly what happened in that accident, they are probably anxiously awaiting your call.

And while we are wantonly slinging statistics, here’s one for you: No Cirrus pilot who has been to the Cirrus Pilot Proficiency Program has been involved in a fatal Cirrus accident. Problem solved!

Sarcastically yours,


Bravo! YES, you got it right! YEA!

Good point about the time in existance of Cirrus. (I can’t even address your point about ANOVA, because I frankly don’t remember enough about it.) There are probably a million variables I’m not considering. I know that just because a statistical analysis assumes something as a random variable doesn’t mean it is and to compare phenomena (like two airplane) requires looking at similar populations if input (pilots). Furthermore, I know statistics NEVER tell you how anything about how safe YOU will be in a particular situation, because most of the factors we assume to be random aren’t really random.

However, for predicting what the accident statistics will be (i.e. just a conclusion on what the data will look like) I believe we have enough information to say they most likely will be worse than the general fleet. Granted, for the reasons you state that doesn’t mean we can draw conclusions about the airplane itself, but it does means there’s something worth looking at in the explanation for the difference.

In fact, it’s those variables that make the Cirrus incompariable statistically to other aircraft that I’d like to discuss and for which I’ve suggested a few possibiliites on this board.