Air Show Flying: Zero Error Margin


The air show arena is a hostile environment. Controlling an aircraft in three dimensions very close to the ground at high closing speeds and high angular rates creates a high pilot workload which, when coupled with the human’s inability to accurately assess closure rates consistently, makes for a high risk of judgement error by the pilot. In air show flying, there is zero error margin.


The old adage of “If you can’t express it in numbers, you do not understand it” is particularly relevant in better appreciating the hazards imposed by low-level air show flying. Analysis of a random sample of 622 air show accidents and incidents worldwide over the 102 year period from 1908 to 2010, confirms the fickleness of human judgement in the low level display arena.

Such analyses do not attempt to apportion blame, but only to tabulate the relevant accident details for the purpose of analysis in an attempt to determine trends in air show accidents. Identifying deficiencies in the air show safety domain and considering ways and means to address any deficiencies will be beneficial to the industry and the air show population at large.

Obviously, pilots do not intentionally crash during an air show performance. The problem is that the environment is hostile and any lapse in concentration by the pilot, misjudgement, poor anticipation, mechanical failure or rash indiscipline, may result in an accident; it’s as simple as that!

In many cases, it is easy to lay blame the pilot, but one must get to understand the underlying reasons why the pilot made a judgement error, a mistake which in most cases, cost lives. The survival instinct of the human is an incredibly strong motivator in avoiding crashes, so what is it that causes the pilot’s physiology to be seduced by peripheral inputs? What is it that convinces the pilot that the selected course of action would offer the best results and survival probability?  What is it that makes an air show pilot, well versed in the threats, “press” outside of practiced and briefed routines?  The answer is yet to be discovered by scientists and human factors engineers. But, how fickle is judgement really? 

Air Show Performance vs. Air Show Practice Accidents

Interestingly, within the scope of the analysis, approximately 70 percent of the accidents occurred during an actual air show while approximately 30 percent occurred during practice sessions. The difference is significant. The most probable causes of this 70:30 split can be attributed to the real world pressures of air show flying; the pressure to perform and to impress is higher during the actual air show than during practice sessions. The training is, unfortunately, by its inherent nature, usually more relaxed.  This emphasizes the tried and trusted military ethos to “train as you intend to fight,” which is directly applicable to air show flying; “you must train as you intend to perform.” Failure to do so introduces unknown physiological variables into the performance routine, which could have bearing on the safety of the spectators and pilots alike.

Contributory factors include the “spectator, peer and organizer pressure” to perform before the public (the attitude that the “show must go on” despite poor environmental conditions of wind, visibility, cloud, rain and density altitude). All these factors can be found as contributory causes to accidents within the population sample of more than 600 air show accidents. Air show organizers are usually loathe to cancel performances with many thousands of spectators having driven for miles and many hours to attend the popular air shows.

And then, of course, there’s the pilots ego: the desire to compete, the desire to impress, the desire to fly the best show. All demonstration pilots will have encountered this emotion and know exactly the feeling being referred to here.  These factors may lead to air show pilots “pressing for maximum effect.” 

Analysis of Air Show Accident Fatalities 

Within the air show accident population sample covered in the analysis, a total of more than 2,113 people were killed or injured in pursuance of air show flying, either as pilots, aircrew, spectators, passengers or even as members of the public totally divorced from the air show itself. Within the accident sample size of 622 accidents, 458 pilots lost their lives while 95 were injured.  Invariably, aircrew in an air show accident perished with the pilots.

One of the major concerns is the high number of spectators killed or injured while attending air shows.  Within the scope of this worldwide analysis, 54 percent (1091) of the victims were spectators, 336 killed and more than 806 spectators injured while attending air shows. The more well known major contributors to such devastating statistics were the Ukrainian Air Force’s Su-27 crash into the spectator enclosures killing 83 and injuring 156 spectators in July 2002; the Frecce Tricolori accident at Ramstein Air Base in Germany in 1988 (69 killed and more than 300 injured); and John Derry’s Sea Vixen accident at Farnborough in 1952 (28 killed and 63 injured).

The imposition of safety regulations by show organizers and the compliance with such regulations by the air show pilots is essential to reduce this unacceptably high percentage of air show fatalities. Failure to do so poses a serious threat to the future of air shows.

The hazards facing spectators at air shows are generally known and regulations are imposed to address spectator safety. However, in the case of the general public not even involved with the air show, there is no elegant solution. The concerns of inhabitants in housing developments around the airfield are well founded.  The area in and around an airfield being used for an air show is at risk to collateral damage in the event of an aircraft crashing outside of the airfield. In the analysis, 72 of the public were killed and 71 were injured, members of the public that just happened to be in the wrong place at the wrong time.

The innocent bystanders, the public killed or injured by collateral air show accident damage, is pertinently illustrated in the case of the Tupolev Tu-144 accident at the Paris Airshow in 1973 in which nine members of the public were killed and sixty injured by crash debris coming down on the village of Goussainville, a few miles from the Le Bourget airport.  In April of 2008, a pilot lost directional control of the Zlin Cmelak “crop sprayer” during takeoff and ploughed into the spectator enclosure, killing one spectator and injuring 14, mostly children, at an air show in Germany.  On September 5, 2010, once again in Germany, a Tiger Moth biplane lost directional control during takeoff at an air show and crashed into the spectator area , killing one and injuring 38.

Analysis of Accident Causal Factors

Analysis of the sample data clearly illustrates that the highest percentage of accidents (73 percent) can be attributed to Man, but, considering the hostile environment of the low-level air show performance arena and the human’s physiological shortcomings, this is not surprising. These catastrophic errors can largely be classified in a small number of areas: flight-into-terrain (29 percent), loss-of-control (22 percent) and mid-air collisions (14 percent). Accidents caused by a mechanical problem (17 percent) or structural failure (7 percent) are comparatively rare.

Of the 183 flight-into-terrain accidents, 65 accidents (36 percent) were in the vertical and 37 (20 percent) were associated with low-level rolling maneuvers.  One of the more disconcerting factors is the relatively high percentage of loop and downline multiple rolls that ended in tragedy. Of the 65 flight-into-terrain fatal accidents that happened during vertical maneuvers, 32 (49 percent) were looping maneuvers and 11 (17 percent) were downline rolls.

So, what is it about the vertical manoeuvres like the loop and  the downline snap roll that gets away from some highly experienced pilots?  Is it poor situational awareness or the amount of energy lost, bearing in mind that energy management in this maneuver is highly dependent on small changes in technique, not to mention time delays?  Do air show pilots really let their energy budget get so close to the edge during an air show? Do pilots realize the extremely high rate of descent when pointing nearly vertically downward ?  The question that must be asked, how many consecutive downline rolls are available from a given height budget? Sadly, the descriptions available from NTSB reports (“Pilot failed to recover.”) do not provide the type of informed observation that are required to conduct a deeper analysis of the question that most air show pilots ask themselves when they hear about this kind of accident: What was it that seduced the pilot into misjudging?

A total of 131 loss-of-control accidents contributed to approximately 22 percent of the air show accidents. The relatively high percentage of accidents of this type also highlights the requirement for above average handling skills among low-level air show pilots. The more well known loss of control accidents included the Russian Tu-144 steep pitch-up which consequently led to loss of control and catastrophic structural failure at the 1973 Paris Air Show; the entry into a spin from a wingover by the De Havilland Mosquito at an air show at Barton, Great Britain in 1996; and the P-63 King Cobra at Biggin Hill Air Fair, also in Great Britain, in 2001. More recently, in August 2010, the pilot of the Swift glider that lost control during the base leg turn for a downwind landing at the Shoreham airshow, miraculously walked away from the wreckage of a near vertical impact.

Mid-air collisions, by their very nature, are unique to formation aerobatics and synchronized pairs displays, occurring principally between members of formation aerobatics teams, military as well as civilian.  Not only do collisions occur in flight, but there are, of course, also cases of individual members of formations colliding while on the ground. Considering the highly dynamic maneuvering of several aircraft flying only a few feet apart at high speeds, it is not surprising that mid-air collisions typically account for 14 percent of air show accidents.

Formation aerobatics most probably represents the pinnacle of low-level display flying hazards; not only does the close proximity of the ground present a hazard, but also the extremely close proximity of the other aircraft in the formation adds significantly to pilot workload and the requirement for situational awareness.  Formation aerobatics requires extremely good reactions, situational awareness, anticipation and fine motor skills from all the pilots in the formation.  Of course, the formation leader is the critical element in controlling the formation.  In 1982, four members of the USAF Thunderbirds aerobatic team were killed when their four T-38 Talons crashed in the Nevada desert while performing a line-abreast loop.  One would have thought that after Frecce Tricolori’s tragic midair at Ramstein in 1988, opposition pass accidents would have become a thing of the past as the display pilots learnt lessons from those that had paid the “ultimate price;” alas, not so! In 2007, we suffered a ‘Ramstein déjà vu’ as two aircraft from the Zelazny Aerobatic Team collided with each other.

In any mechanical system, there will always be a certain failure rate, the mean-time-between-failure.  Machine’s contribution to air show accidents, specifically mechanical, was 17 percent (109 failures) of which 65 (58 percent) were engine failures while 34 structural failures also contributed to mechanical failures.  Given the high demands put on the airframe and engine during an aggressive aerobatic routine, the comparatively small number of accidents caused by mechanical failures is both somewhat surprising and is possibly a testament to the strong emphasis that most air show pilots place on maintenance.

Structural failures often result in some of the most spectacular crashes. John Derry’s DH-110 in-flight structural break-up at Farnborough 1952, and the in-flight wing structural failure of the F-117 in Baltimore in 1997 are cases in point. There is no statistical evidence to substantiate the suggestion that vintage aircraft are more likely to experience structural failure, despite their age.  On the contrary, four of the cited examples were current operational aircraft at the time of the accident and three were ex-service or vintage aircraft types.

And then, of course, there is the element of Medium, the environment.  In aviation, Medium will always contribute a certain percentage to accident causal factors, typically in the form of bird strikes and weather-related elements such as density altitude, cloud base, visibility and wake turbulence, to list just a few.  In the sample analysis, however, Medium’s direct contribution to display accidents is surprisingly low at only three percent. However, this does not reveal the true picture.

Weather contributed directly to at least three and indirectly to at least eight air show accidents, accidents in which pilots were deceived into “pressing” their performances, trying to squeeze their displays in below the cloud base in conditions which should in most cases, have resulted in cancellation of the air show. The B-58 Hustler barrel-roll against a low cloud ceiling at Paris Airshow 1961, the A-10 Thunderbolt attempt to squeeze the loop in below the low cloud ceiling, also at Paris Airshow, but in 1977, and the synchro-pair of MiG-29s at RIAT 1993 that collided on exit from “light” clouds while carrying out the “lost wingman” procedure, serve as prime examples as to the hazards of conducting aerobatics at low-level with cloud present.  There are many other examples where adverse weather was the indirect cause of an air show accident.

In several cases, strong crosswinds forced pilots to approach or cross the crowd line, requiring the pilots to take aggressive action to avoid the wrath of the safety directors or of overflying prohibited overflight areas. Density altitude issues caused pilots to miss the “energy gates.” Orographic turbulence may have contributed to formation team members colliding with each other. Etc.  Considering that air show flying is essentially a VFR exercise, persistence and logic in flying displays at air shows in adverse weather conditions is questionable.  The existence of a practiced “bad-weather” sequence in any display repertoire is essential to eliminate such accident categories. The price for non-compliance can prove fatal for not only the pilot, but also the spectators. Even at the relatively low percentage contribution of three percent, the number of accidents caused by weather constraints is excessive. It is one of the few areas of air show accidents that can be theoretically reduced to zero, but that would require the consistent demonstration of fine judgement and courage by both air show pilots and event organizers to recognize when conditions have deteriorated and make the hard decision to cancel subsequent flying.

The Royal Jordanian Air Force aerobatic display at the Flanders Fly-In in Belgium in 1997 was flown in such heavy rain and strong winds that many of the few remaining spectators took shelter in a First Aid tent, not comprehending the impact of the strong crosswinds on the pilot’s ability to maintain the showline. Several spectators were killed when one of the team’s Extra 300 aircraft hit the tent and the people who had sought shelter under it.    

Man’s Role

The bottom line is that not all pilots are “good sticks;” they may fly safely and accurately during their “day job,” but it takes a different type of pilot and attitude to fly an aircraft aerobatically at low-level.  Conversely, some exceptional air show pilots would not survive long in a more conventional type of flying environment. These statistics leave one in no doubt as to the fallibility and weakness of Man in the low-level air show environment. Armed with these facts, it is essential for Man to focus on his weaknesses and develop counters for them to improve survivability indices.

Certainly one of the most significant observations emanating from the analysis is that the pilots were all generally highly experienced, professional aviators; flying experience ranged from 2,000 to 20,000 flying hours and, in several cases, many years flying on the air show circuit. In most cases, the accident investigation reports concluded that the pilots had erred in some way or another. The important point to be made here is that none of the errors were committed intentionally. What must be asked is what made the pilot respond in the manner that caused the particular accident.  Why was that particular course of action pursued?  What was it that drove the pilot to think that his incorrect chosen course of action would be successful?  What distractions in the cockpit made the pilot select the incorrect course of action?  Did the pilot experience physiological problems? Were his sensors overloaded with peripheral cues and rate information, or did he just decide to “take a chance,” thinking that it would be alright?  Was it an educated decision or was it just taking a chance?

We will never know the answers to these questions.  Accident investigation boards make conclusions based on a given set of accident evidence; they are never privy to those vital cues, information or terrors that the pilot faced at the time of decision making. In most cases, they can never know exactly what happened in the cockpit. The bottom line is, in most cases, so-called “pilot error” has an extenuation factor that the accident investigation board do not necessarily discover during the investigation.  Not that this stops the accident board making the standard “clique finding” of ‘”pilot error.”

Based on the legacy air show accidents then, air show organizers and aviation authorities world wide, both military and civil, have had no other alternative but to introduce and impose regulations governing the safety of the spectators and public firstly, and the pilots secondly.  Air show accidents are not a new phenomenon; the statistical evidence bears witness to vast number of accidents that have occurred in the past and statistically, the number of airshow accidents annually at air shows worldwide has been increasing. On the good news side, the number of spectators and members of the public that have been killed or injured in Western countries has not shown any significant increase which implies that the imposed regulations are, in most cases, successful.  


As an international air show community, we seem to have reached a plateau in our ability to reduce accidents and incidents at airshows. The question is: Can we afford, then, to just continue and accept an average of 23 accidents/incidents per annum? Can we afford to just accept what the dice have dealt with the associated loss of life? Are we doing enough worldwide to reduce the number of air show accidents/incidents? Based on the fickleness of human judgement in the low level air show environment, is it realistic to believe that zero accidents are possible in a given year?  If not, what is an acceptable loss rate?  Should one even consider anything else than a target of zero accidents per annum?  These and many other questions need to be answered if we, as the air show community, can deliver on maximum safety.

Previous articleAerobatic Sequences: Construction, Revision and Peer Review
Next articleEmbracing Social Media
Des Barker is a former pilot and general officer in the South African Air Force who also prepars an annual summary and analysis of air show accidents around the globe.