A first look at the airplane, the operator, the runway, the weather record, and how a 767 stops
Analysis by Attorney and Former Commercial Airline Captain and Chief Pilot Christopher Rusing; and by Attorney and Commercial Airline Pilot Brendan Keegan
The attorneys at Aviation Law Group, PLLC, are shocked and saddened by the events that unfolded on Sunday afternoon on a Miami street outside of Miami International Airport runway. Five people lost their lives and five others were injured, three seriously when a cargo aircraft overran the airport boundary while attempting to land.We grieve with the families who lost loved ones today, and those whose family members were injured.
On Sunday afternoon, September 6, 2026, a Boeing 767-300 freighter operated by 21 Air for Amazon Prime Air overran the end of Runway 30 at Miami International Airport (MIA) after landing. The airplane crossed the airport perimeter road and NW 67th Avenue, struck several vehicles, and came to rest roughly 2,000 feet beyond the pavement, where it caught fire. Five people were killed and five more were injured, three of them critically. The NTSB has launched a go-team, and Chair Jennifer Homendy is the on-scene board member.
Our firm in investigating the accident as be do after every major accident: examine the public record, reviewing the weather data minute by minute, and analyzing at the airplane and the runway with a pilot’s eye. What follows is the information we have been able to assemble that has not yet appeared in most news coverage, along with an explanation of why each piece matters to the investigation that is now underway.
The Airplane
The aircraft was N1997A, a Boeing 767-33A(ER) converted freighter, manufacturer’s serial number 27310, powered by two General Electric CF6-80C2B6F engines. It first flew on June 17, 1994, and was delivered new to the Belgian charter carrier Sobelair. Over the next twenty years it passed through Vietnam Airlines, Air Europa, Kenya Airways, and the Russian carrier Nordwind, with two periods in storage. In late 2015 it was ferried to Tel Aviv, where Israel Aerospace Industries’ Bedek division converted it from a passenger airplane to a freighter. It entered cargo service with Atlas Air in August 2016 as one of the first aircraft painted in Amazon’s Prime Air colors, and was widely photographed under the name “Amazon One.” In January 2025 it moved from Atlas to 21 Air, its seventh operator. At the time of the accident the airframe was 32 years old.
According to the FAA registry, the registered owner is Andromeda Leasing II LLC, with an address at 2000 Westchester Avenue in Purchase, New York. That is the headquarters address of Atlas Air Worldwide. So, before anyone reaches the question of maintenance, there are already three distinct companies in the chain: an Atlas-affiliated lessor that owns the airplane, Amazon, which by published accounts controls and supplies the 767-300 fleet flown on its behalf, and 21 Air, which holds the operating certificate and employs the crew. Who actually performed heavy and line maintenance on N1997A, who managed its brake and thrust reverser components, and what the lease required of each party are questions the investigation will seek to answer.
The Operator
21 Air LLC is a Part 121 cargo carrier headquartered in Greensboro, North Carolina, with operating bases at Piedmont Triad International and at Miami International. It was formed in 2014 as a virtual airline flying under Dynamic Airways’ certificate and later obtained its own. It is owned by Avia Acquisitions LLC. The Canadian cargo carrier Cargojet held a 25 percent stake until it divested in April 2026. The fleet has roughly doubled in two years, from eight aircraft to fifteen or sixteen, and now includes 767-300 converted freighters flown for Amazon, 767-200 freighters flown for DHL, and 757-200 freighters. Amazon flying began in late 2024. The company’s chief executive departed in February 2026 when his contract expired, and trade reporting through the spring described leadership turnover and delays in bringing newly acquired 757s through FAA certification.
None of that is causal. Investigators routinely look at rapid fleet growth, management change, and regulatory oversight because those conditions shape training, scheduling, and maintenance capacity. We mention them because they will be in the docket, and readers should understand why.
The Weather, Minute by Minute
Investigators with the NTSB will focus on the weather as thunderstorms near the airport may have played a significant role the fast-changing weather conditions the pilots may have encountered as they commenced their approach for landing.
At 12:53 pm local time (1653Z), an hour before the accident, Miami reported wind from 260 degrees at 5 knots, with cumulonimbus clouds to the south and southwest moving east. At 1:13 pm a special observation recorded that a thunderstorm had begun over the field at 1:12 pm, with variable light winds. The 1:53 pm observation, taken within five minutes of the landing, is the one that matters. Verbatim:
METAR KMIA 061753Z 19017G26KT 10SM TS SCT020CB BKN048 BKN150 BKN250 30/22 A2994 RMK AO2 PK WND 18026/1752 WSHFT 1737 SLP137 TSB12 OCNL LTGICCG E-S TS E-S MOV E CB DSNT NW
In plain language: wind from 190 degrees at 17 knots, gusting to 26; a thunderstorm on the field with occasional lightning to the east and south; a wind shift recorded at 1:37 pm, which is the signature of a gust front passing over the airport; and, in the remarks section, a peak wind of 26 knots from 180 degrees recorded at 1:52 pm. Twenty minutes after the accident the wind was down to 12 gusting 17, and an hour later it was calm.
The aircraft may have landed with a tailwind. 10 knots is the standard tailwind limitation for landing on the Boeing 767, and on most Boeing and other transport-category jets. We are aware of at least one major U.S. 767 operator whose manuals impose exactly that limit.
A flight crew does not necessarily know the peak wind. The ATIS broadcast carries the two-minute average wind from the airport sensor. A tower wind check gives current direction, speed, and gust. The peak wind appears only in the remarks section of the written METAR, only when it exceeds 25 knots, and only after the observation is compiled. The one-minute sensor data that would show exactly what the wind was doing at the moment of touchdown exists, but it lives in a federal archive, not in the cockpit.
One further point about the weather. Which runways were in use, why, and what wind information was passed to this crew on final are questions for the air traffic control recordings, which the NTSB will obtain and seek to answer questions why it was not in use despite the weather indicating the opposite direction landing was favored.
Runway 30
Media coverage has described Runway 30 as 9,360 feet long. That is the length of the pavement. It is not the landing distance available. Runway 30 has a displaced threshold of 945 feet, meaning the touchdown point is set well down the pavement, and the FAA’s published landing distance available for Runway 30 is 7,913 feet. It is the shortest declared landing distance of any runway at Miami. By comparison, Runway 9/27 is 13,016 feet and Runway 8R/26L is 10,506 feet.
Beyond the end of Runway 30 lies the airport perimeter road, then NW 67th Avenue, a public street with traffic on it, then the parking area and field where the airplane stopped. The FAA design standard for a runway safety area at a runway of this class is 1,000 feet beyond the runway end.
The people who passed on Sunday were, by every account so far, on or near that public road. In the history of U.S. airline runway overruns, deaths on the ground outside the airport fence are rare. The closest precedent is Southwest Airlines Flight 1248 at Chicago Midway in December 2005, which went through the airport fence onto Central Avenue and a six-year-old boy in a car lost his life. The relationship between the airport’s safety area, the road, and the fatalities on Sunday will be a central issue in this investigation.
How a 767 Stops, and Why the Numbers are Smaller Than They Look
A transport jet’s certified landing distance under 14 CFR § 25.125 is measured from 50 feet above the threshold to a full stop on a dry runway with maximum wheel braking and no credit for reverse thrust. For an airline to dispatch a flight, 14 CFR § 121.195 requires that the airplane be able to stop within 60 percent of the available runway at the destination, and that figure is increased by 15 percent if the runway is expected to be wet. For a 767-300ER at its maximum landing weight of 320,000 pounds, Boeing’s airport planning data put the resulting field length in the range of 5,900 to 6,000 feet dry and about 6,900 feet wet. The actual unfactored stopping distance is considerably shorter. The difference is the safety margin, and the margin is the point. Runway 30’s 7,913 feet is adequate for a 767 on paper. The question is always what happens to the margin when conditions change.
Since 2006 the FAA has directed airlines, through Safety Alert for Operators 06012 and the later Takeoff and Landing Performance Assessment guidance in SAFO 19001, to recompute landing distance at the time of arrival using the actual conditions, including runway condition, wind, weight, and planned use of autobrakes and reverse thrust, and to add a margin of at least 15 percent. What number this crew computed, if any, and with what inputs will be in the record.
The FAA’s own advisory circular on runway overrun prevention, AC 91-79B, gives a rule of thumb: a tailwind increases landing distance by about 21 percent for the first 10 knots. A 10 percent increase in approach speed increases landing distance by about 20 percent. Every 10 feet above the standard 50 foot threshold crossing height adds about 200 feet. And a two-second delay in getting the deceleration devices working at 118 knots costs almost 400 feet; at the higher speeds a heavy 767 lands at, the cost is greater. These factors compound. They do not add.
The Stabilized Approach
Pilots and regulators describe a landing approach as “stabilized” when, by 1,000 feet above the ground in instrument conditions or 500 feet in visual conditions, the airplane is on the correct lateral and vertical path, in its landing configuration, at the target speed within a narrow band, descending at no more than 1,000 feet per minute, with the engines at an appropriate thrust setting, and with only small corrections needed. The criteria come from the Flight Safety Foundation’s
Approach and Landing Accident Reduction work and are embedded in FAA guidance and in every airline’s operating manual. They exist because unstable approaches are the leading precursor to runway excursions worldwide.
Weather like Sunday’s is the enemy of a stabilized approach. Gusts require higher approach speeds by design; airline procedures add a gust increment to the reference speed, up to a certain maximum, in many cases about 20 knots. Thunderstorms can produce wind shear that can change airspeed and sink rate in seconds and have been the cause far too many airline accidents in the past.
The NTSB will examine the flight data recorder (FDR). It will show airspeed, glideslope deviation, engine power, pitch attitude, and descent rate second by second down the approach. If weather was a factor in this accident, the FDR will show it.
The Touchdown Zone and the Go-Around
Airline procedures define the touchdown zone as the first 3,000 feet of the runway or the first third, whichever is less. On Runway 30, with 7,913 feet available, the first third is roughly 2,600 feet. The standard rule, written into the Boeing Flight Crew Training Manual and into airline manuals across the industry, is that if the airplane has not touched down within the touchdown zone, the crew goes around. A go-around is a normal maneuver that crews train and practice.
The industry also knows how hard that decision is in practice. The Flight Safety Foundation’s 2017 study of go-around decision-making found that only about 3 percent of unstable approaches end in a go-around. On a day like Sunday there is an added trap: going around means climbing back into the thunderstorm that created the problem. The option is least attractive at the moment it is most needed. That is a systemic feature of convective weather at busy airports, and it is the reason the industry keeps studying it.
Unofficial ADS-B tracking data suggests the airplane was still moving at about 177 knots over the ground somewhere along the runway and at roughly 110 knots as it left the pavement. Those figures are preliminary and their position along the runway is uncertain. The investigators will fix the actual touchdown point from the recorder and from the tire marks.
What Videos Appear to Show
Videos circulating since Sunday afternoon on social media appear to show the airplane touching down roughly halfway along Runway 30, with the right main landing gear on the pavement and the left main gear still in the air. It does not appear to show the thrust reverser sleeves opening, and it does not appear to show the wing spoilers rising.
The NTSB will review the video evidence to assist in its investigations. The 767 aircraft’s systems play an important role is stopping the aircraft when landing. Thrust reversers and spoilers play an important role is slowing the aircraft along with its brakes. The thrust reversers and spoilers will not work if the airplane’s systems thinks it is still flying. The 767 decides whether it is in the air or on the ground using angle sensors on each main landing gear truck, along with a sensor on the nose strut. Both main trucks must be in the ground, in an un-tilted position, and the two sensors must agree, before the airplane’s logic switches to ground mode. Until that happens, the automatic spoilers will not deploy, and the thrust reversers will not deploy even if the pilots have pulled the reverse levers. If the both main gear were not on the ground , the airplane, as designed, would have treated itself as still flying, preventing the additional assistance in slowing the aircraft after landing.
In a previous incident on May 22, 1997, a 767 landing in Newark, New Jersey, touched down in a gusty crosswind of 17 gusting 29 knots after a windshear alert five seconds before touchdown. The flight data recorder showed the sequence precisely. In the NTSB’s words: “The right and then the left main landing gear tilt switches transitioned to the ground position, and the speed brake handle deployed.” The spoilers came up only after both trucks were down. Two seconds later the nose gear hit hard, and the airplane’s pitch leveled momentarily; the recorder showed “first, the left, and then the right main landing gear tilt switches transitioning back to the flight mode, and the speed brake handle returning to the stowed position.” One truck lifting was enough to stow the spoilers.
In the 1997 incident, the NTSB made a finding that should be read carefully by anyone thinking about Sunday’s accident: “once the speed brake lever returned to the stowed position and the speed brake panels on the wings retracted, a further return to the ground position of the tilt switches would not re-deploy the speed brakes. They would have to be manually re-deployed.” In other words, a skip or a rocking touchdown does not merely delay the spoilers. It can cancel their automatic deployment for the remainder of the landing roll unless a pilot notices and pulls the lever by hand.
The NTSB will also examine the Minimum Equipment List to see what equipment may not have been operative on this particulary 767. Airlines may legally fly a 767 with one thrust reverser or the automatic spoiler system deferred and inoperative, with a landing performance penalty and specific crew procedures. Whether N1997A was carrying any such deferral is a maintenance-records question, and the answer will be in the docket.
Where This Goes From Here
The NTSB will recover the flight data and cockpit voice recorders, document the tire marks on Runway 30, examine the position of the reverser sleeves and spoiler actuators in the wreckage, obtain the air traffic control recordings and the one-minute weather sensor data, and pull the maintenance records for N1997A back through its conversion. A preliminary report typically follows within a few weeks and will contain none of the analysis. The full investigation will take a year or more. The families of the people who were killed and injured on NW 67th Avenue on Sunday will learn what happened from that process, and they should be wary of anyone who tells them the answer before then.
About Our Experience
The Aviation Law Group, PLLC (ALG) is a firm devoted exclusively to aviation accident litigation.
Our attorneys are all licensed pilots and some are also former aircraft mechanics. Collectively, our lawyers hold thousands of hours of flight and instructional time and two of our attorneys have flown or continue to fly in international airline operations. We know the culture, the equipment, and the realities of major airline operational pressures firsthand.
ALG maintains offices in South Florida, Washington State, and Hawaii with attorneys licensed in Florida, Washington, Hawaii, Alaska, California, Texas, New Mexico, and Vermont, and through association with local counsel, represent clients in all 50 states and internationally. If your family was affected by the Amazon Prime Air crash and you would like to understand what happened and what your options are, we would be honored to talk with you in a confidential, no-obligation consultation.
Contact info@aviationlawgroup.com or call 206-464-1166