A runway overrun may be the end of a sequence that began minutes earlier. During that sequence, safety margins can narrow through changes in speed, height, wind, touchdown point, braking performance and the decisions made as each condition develops. The accident at Miami International Airport invites us to examine that progression, even though investigators have not yet determined how it unfolded in this case.
On 6 September 2026, at approximately 2:00 p.m. local time, 21 Air Flight 7598, a Boeing 767-300 freighter operating for Amazon Air, overran Runway 30 at Miami International Airport after arriving from San Juan, Puerto Rico. The aircraft crossed the airport boundary, struck vehicles beyond the runway environment and caught fire. Initial reporting identified five fatalities and five injuries. The two pilots survived, although they were injured.[i]
The five people who died were occupants of vehicles struck beyond the runway environment, including a van carrying cleaning-company employees. That fact changes the scope of the safety discussion. The consequences were not confined to the aircraft or airport movement area; they reached people who had no part in the flight.
The investigation is at an early stage. The National Transportation Safety Board dispatched a team and recovered the flight data and cockpit voice recorders. Investigators will examine the aircraft, the recorders, weather, runway conditions, landing performance, crew actions and operating procedures before determining probable cause. Until then, incomplete flight-tracking data and online video can raise questions, but they cannot answer them.
The responsible question is therefore broader than who may have made an error. How can a landing progressively run out of safe options? Five connected stages help organise that inquiry: arriving with manageable energy, deciding whether to continue, controlling the touchdown, recognising inadequate deceleration and limiting the consequences beyond the runway.
Arriving With Manageable Energy
Every landing is an exercise in energy management. An aircraft approaching a runway carries kinetic energy through its speed and potential energy through its height. Its mass, vertical speed, configuration, wind and atmospheric conditions affect how that energy must be managed. The crew must bring these elements into balance so that the aircraft reaches the runway at an appropriate speed, attitude and point. The runway then supplies a finite distance in which the remaining energy must be dissipated.
Excess height must be lost and excess speed reduced. A tailwind increases groundspeed. A prolonged flare consumes pavement. Delayed ground-spoiler deployment, wheel braking or reverse thrust leaves less distance for those systems to act. Rain or contamination can reduce the deceleration available. None of these factors has yet been established as causal in Miami; together they explain why an approach must be assessed as a developing energy state rather than as a path that happens to terminate at a runway.
Pilots and operators address this through stabilised-approach criteria. These commonly include airspeed, descent rate, flight path, aircraft configuration, thrust setting, checklist completion and position relative to the runway. When the prescribed conditions are not met by a defined altitude, a go-around is normally required.
Numerical compliance alone, however, does not guarantee a safe outcome. An aircraft may remain within individual tolerances while the combined energy picture deteriorates. The crew must continue asking whether the aircraft is converging on a safe touchdown with sufficient runway remaining. Airspeed must be controlled, the intended touchdown zone must remain attainable, and changing wind or runway conditions must remain compatible with the landing-performance assumptions. A stabilised approach is a continuing assessment, not a status acquired once and retained automatically.
The Decision to Continue
The next stage is the decision to continue or discontinue the approach. A go-around is sometimes experienced as a failed landing, even though it is an established safety manoeuvre. Properly understood, it records a judgment that the conditions or margins needed for landing are no longer sufficiently assured.
Continuation bias can make that judgment harder. Most approaches end in landings, so each completed step creates momentum toward the runway. The descent has been flown, landing clearance received and the aircraft configured. The destination is close, and operational expectations favour arrival. As the aircraft nears the runway, abandoning the attempt may feel increasingly disruptive even when a go-around remains available.
Operators must therefore make go-arounds psychologically acceptable as well as procedurally clear. Training should expose crews to late-developing situations in which discontinuing the approach is the expected decision. Captains should encourage timely challenges from first officers, and standard calls should leave little room for ambiguity. Flight-data monitoring can also reveal whether crews are routinely continuing approaches with diminishing margins.
The decision may not arise at one dramatic moment. The aircraft may be high, speed may not reduce as expected, a gust may disturb the flight path, or the intended touchdown area may begin to pass beneath the aircraft. Each development requires a renewed assessment of whether landing remains the safer course. Investigators will determine whether such decision points arose at Miami and how the crew responded. Publicly available images may inform questions for investigators, but they are not findings.
Where the Aircraft Touches Down
Published runway length can create a false sense of comfort. During landing, the relevant measurement is the runway remaining after the aircraft is firmly on the ground and capable of effective deceleration.
The sequence is simple: approach, flare, touchdown, deceleration and stop. Every metre used during an extended flare or before effective ground contact is unavailable for braking. If touchdown occurs significantly beyond the planned zone, the landing-distance calculation made before arrival may no longer describe the situation the crew faces. Touchdown-point control is therefore a primary safety requirement, not merely a measure of handling finesse. A positive touchdown in the intended area may preserve more margin than a prolonged attempt to achieve exceptional smoothness.
Touchdown also marks the transition from aerodynamic flight to ground deceleration. Weight must settle onto the wheels. Ground spoilers reduce lift and help transfer weight to the landing gear, allowing the wheel brakes to work effectively. Reverse thrust contributes most during the higher-speed portion of the landing roll. If the aircraft floats, bounces, touches down in an unexpected attitude or does not achieve effective weight-on-wheels promptly, the deceleration sequence may be delayed while the available runway continues to decrease.
Flightradar24 reported that ADS-B data showed the Miami aircraft travelling at approximately 112 knots as it left the usable runway.[ii] That figure indicates the severity of the overrun but does not establish touchdown speed, braking effectiveness, aircraft configuration or crew actions. It describes a tracked point near the runway end, not the earlier sequence that produced it. Those answers must come from the official investigation.
If touchdown has not occurred within the operator’s permitted zone, a go-around may still be required. Once the wheels are down, the judgment becomes more complicated. Whether a balked landing remains feasible depends on the aircraft’s condition and configuration and on approved procedures. After spoilers, reverse thrust or heavy braking have been committed, attempting to fly again may introduce additional risk. The options narrow quickly after ground contact, which is why decisions made earlier in the approach carry such importance.
Recognising Inadequate Deceleration
Once firmly on the runway, the crew must determine whether the aircraft is slowing at the expected rate. At 120 knots an aircraft travels roughly one kilometre in 16 seconds. A delay of only a few seconds can consume hundreds of metres.
Effective deceleration depends on touchdown speed and point, runway slope and surface condition, wind, wheel braking, antiskid performance, spoiler deployment, reverse thrust and aircraft mass. A modest degradation in one area may be manageable. Several occurring together can remove the available stopping margin quickly.
Weather belongs within this calculation. Reported conditions at Miami included thunderstorms and winds around 17 knots gusting to 26 knots.[iii] Wind speed alone does not identify the headwind, tailwind or crosswind component experienced on Runway 30, and the reported conditions do not establish that weather caused the accident. They do illustrate the uncertainty associated with convection. Thunderstorms can produce rapidly changing winds, turbulence, windshear and intense rain within the duration of a single approach, while also increasing crew workload.
A sound pre-landing assessment must therefore consider more than whether a reported value is within a limit. It must preserve margin for changing conditions and for the possibility that actual braking performance will differ from the calculation. Crews also need clear cues for recognising inadequate deceleration. If the expected relationship between speed and runway position is not developing, immediate recognition supports maximum braking, directional control, timely warning to air traffic control and preparation for an overrun.
What Lies Beyond the Runway
The final stage lies outside the cockpit. Aviation relies on layered defences because no operational system is perfect. An approach can become unstable, a go-around opportunity can be missed, touchdown can occur long, braking performance can disappoint or equipment can malfunction. Airport design must therefore consider what happens when an aircraft passes the runway end.
A runway safety area is a clear, graded space intended to reduce the consequences of an undershoot, overrun or excursion and to provide access for emergency responders. Where the full safety area cannot practicably be provided, an Engineered Materials Arresting System may offer an alternative. EMAS uses engineered, crushable material that deforms beneath an aircraft’s wheels and dissipates kinetic energy. According to the Federal Aviation Administration, a standard installation is designed to stop the runway’s critical aircraft when it enters the bed at 70 knots or less.[iv] That is a site-specific design case, not a promise that every aircraft at every speed will be stopped.
The reported 112-knot tracking figure cannot be compared directly with the FAA’s 70-knot design case to decide what a hypothetical arresting bed might have achieved at Miami. Entry speed, aircraft mass, landing-gear characteristics, trajectory and the dimensions and location of an installation would all matter. Only a site-specific engineering analysis could determine whether a different runway-end treatment might have altered the outcome.
EMAS is not a substitute for a stable approach, a timely go-around, touchdown-zone discipline or effective braking. It is a final protective layer where it can be appropriately installed. The same systems analysis must also include roads, buildings, vehicles and public areas near extended runway centrelines. As airport activity and surrounding development change, the exposure of people and infrastructure deserves periodic reassessment.
Miami has experienced another cargo-aircraft accident with consequences beyond the airport operating area. In August 1997, Fine Air Flight 101, a DC-8 freighter, crashed shortly after takeoff following improper cargo loading that produced an uncontrollable aft centre of gravity. Four people aboard and a motorist on the ground were killed.[v] The causes of that accident and the 2026 overrun cannot presently be compared. What they share is a consequence: people on the ground were exposed to an aircraft accident originating within the airport environment.
That consequence turns runway-end planning into a public-safety concern. A complete systems review may also consider how responsibilities were divided among the aircraft operator, its commercial customer, the airport and other organisations supporting the flight. Raising that question does not presume that any contractual or organisational relationship contributed to the accident. It recognises that investigators must understand the operating system as well as the actions taken in the cockpit.
Questions Worth Carrying Forward
The investigation must establish how the aircraft arrived over the runway, where effective ground contact occurred, how much stopping distance remained, why substantial energy persisted near the runway end and what protection existed beyond it. These questions allow individual decisions to be examined without treating the cockpit as the only place where safety is created.
A runway overrun seldom begins during the final metres of pavement. It develops as margins change and opportunities to recover them become fewer. Safe operations depend on recognising that progression early. Safe airport design depends on limiting the consequences when operational defences do not succeed.
The most useful lesson to take from Miami, while the evidence is still being assembled, is the need to examine the complete chain. The approach must preserve a manageable energy state. The decision to continue must remain valid. Touchdown must occur where stopping is achievable. Deceleration must develop as expected. Beyond the runway, the environment must offer as much protection as reasonably practicable.
The final report will determine what happened in Miami. Until then, the professional obligation is to preserve uncertainty about the cause while asking precise questions about the defences that should prevent an overrun, or reduce its consequences, on the next flight.
Author Note
The Miami investigation remains preliminary. This article reflects information available in the days immediately following the accident, and facts may change as investigators examine the recorders, aircraft systems, weather, runway conditions, crew actions and organisational factors. It was prompted in part by professional discussion on LinkedIn concerning the operational and runway-design lessons raised by the accident; the analysis and wording presented here are the author’s own.
Sources
[i] Associated Press, Federal investigators probe Amazon cargo jet’s fiery runway crash that killed 5 in Miami, 7 September 2026. https://apnews.com/article/d5118998e880f380a3fd8786e35fd9bb
[ii] Flightradar24, 21 Air 767 overruns runway on landing in Miami, updated 6 September 2026. https://www.flightradar24.com/blog/flight-tracking-news/major-incident/21-air-767-overruns-runway-on-landing-in-miami/
[iii] Ibid
[iv] Federal Aviation Administration, Engineered Materials Arresting Systems. https://www.faa.gov/airports/engineering/incursions_excursions/emas
[v] National Transportation Safety Board, Fine Airlines Flight 101, investigation DCA97MA059. https://www.ntsb.gov/investigations/Pages/DCA97MA059.aspx
