An airliner runs out of fuel when the fuel on board is less than the crew believes, and nothing tells them in time. In the accidents below, that gap came from a unit conversion done with the wrong factor, from a fuel line that split over the ocean, and from a crew that could no longer act because they had lost consciousness. Two of these flights glided to a runway and everyone survived. One did not.
This guide explains how an airliner can end up with empty tanks, using what the investigators found, and what changed after each case. The videos are on the Fuel exhaustion page.
How an airliner can run out of fuel
An airliner leaves with the fuel for its trip plus reserves, and the crew checks the fuel against the flight plan as they go. To end up with empty tanks, one of three things has to happen:
- Too little fuel is loaded, and every check of the figure repeats the same mistake.
- Fuel is lost in flight, through a leak, faster than the crew realises.
- The flight goes on far longer than planned, because of holding, a diversion, or because nobody on board is able to land the aircraft.
Each case below is one of these. None of them comes down to a single error. In each one the fuel problem got past several checks, which is the pattern the Swiss cheese model describes.
Fuel exhaustion vs fuel starvation
Investigators separate two kinds of fuel event. The Australian Transport Safety Bureau’s 2013 report Starved and exhausted defines them like this:
- Fuel exhaustion: no usable fuel is left on board to feed the engines.
- Fuel starvation: there is fuel on board, but it is not reaching the engines, for example because the wrong tank is selected or a valve is closed.
All three cases in this guide with event pages are fuel exhaustion: the tanks really were empty. They are grouped on the fuel exhaustion topic page. Exhaustion usually comes from a major fuelling error or from a series of small slips in planning and in-flight checks.
The Gimli Glider: a unit mix-up and a 767 glide
On 23 July 1983, Air Canada Flight 143, the Gimli Glider, was a new Boeing 767-233, registration C-GAUN, flying from Montreal to Edmonton via Ottawa with 61 passengers and 8 crew. Over Red Lake, Ontario, both engines stopped for lack of fuel.
Why the fuel ran out. The 767’s fuel gauges were blank because of a faulty fuel quantity processor, and the aircraft was dispatched without them. The Board of Inquiry found this was contrary to the minimum equipment list. With no gauges, the tanks were measured by hand with a dripstick. The 767 was the only aircraft in Air Canada’s fleet that measured fuel in kilograms, so the litres measured had to be converted. The figure used was 1.77, the number of pounds in a litre of fuel, instead of about 0.8 kilograms per litre. The flight needed 22,300 kg, and it left with roughly 22,300 pounds, about half. The same calculation was repeated in Ottawa. The flight management computer showed plenty of fuel because it had been given the wrong starting figure. The conversion is worked through in Air Canada Flight 143: The Pounds-for-Kilograms Fuel Error.
The glide. With both engines stopped, the colour displays went dark and a ram air turbine supplied just enough hydraulic power to move the controls. First Officer Maurice Quintal worked out that they could not reach Winnipeg, and at 20:32 Central Daylight Time the crew turned for Gimli, a disused air force base. Captain Robert Pearson, an experienced glider pilot, used a sideslip to lose height without gaining speed. The nose gear collapsed on landing and the aircraft stopped short of people at a drag-racing event on part of the old runway. All 69 people on board survived. About ten were slightly hurt on the evacuation slides.
What the inquiry found. The Board of Inquiry, led by Justice George Lockwood, reported in April 1985. It found errors by the crew and by maintenance, and it placed much of the responsibility on Air Canada: a metric aircraft brought into a fleet that worked in pounds, nobody clearly given the job of calculating fuel, too few spare processors, and too little training. It recommended one unit of measurement across the fleet, dripsticks that read in kilograms, and independent checks of every fuel load.
The aircraft is on the Boeing 767 page, and the landing is one of the stories in When the pilots saved the plane.
Air Transat 236: a leak over the Atlantic and a 65-mile glide
On 24 August 2001, Air Transat Flight 236, an Airbus A330-243 registered C-GITS, was flying from Toronto to Lisbon with 293 passengers and 13 crew.
Why the fuel ran out. Days before the flight, mechanics had replaced the right engine with a loaned unit in an older configuration. The service bulletin was not fully incorporated, which left a hydraulic line in contact with a fuel line. In flight, the hydraulic line rubbed through the fuel line. According to the FAA’s summary of the Portuguese final report, the leak began 3 hours 46 minutes into the flight and was not detected by the crew. The leak was estimated at 12 to 15 tonnes an hour, several times what the engines were burning.
Why the crew did not recognise it. First the right engine showed unusual oil readings, and then a fuel imbalance advisory appeared. The crew opened the crossfeed to balance the tanks. That sent fuel from the left tanks to the leaking right side. They believed the readings were a computer fault, and they did not use the fuel leak procedure. One hour and seven minutes after the leak began, they diverted towards Lajes in the Azores.
The glide. The right engine flamed out about 150 nautical miles from Lajes. The left engine followed when the aircraft was about 65 nautical miles out at about 34,500 feet. With a ram air turbine for power, Captain Robert Piché flew a turn and S-turns to lose height and landed on runway 33 at 06:45 UTC. Eight of the 12 tyres burst. Everyone survived. Sixteen passengers and two cabin crew were hurt during the evacuation. See Air Transat Flight 236: How a Fuel Leak Stopped Both Engines.
What the report found. The Portuguese investigation (GPIAA) did not name a single probable cause. It listed 12 findings as to causes and contributing factors. They included the engine-change process that missed the configuration difference, the lead technician relying on verbal advice instead of the service bulletin, and the crew’s lack of training for leaks that show up as an imbalance. The aircraft type is on the Airbus A330 page.
Helios 522: when the engines stopped on a plane with no one flying it
Helios Airways Flight 522 is a different kind of fuel exhaustion. On 14 August 2005 the Boeing 737-31S, 5B-DBY, took off from Larnaca with its pressurisation mode selector still in manual after a pressure test the night before. The cabin never pressurised, and the crew were overcome by hypoxia. The cause of that, and why the warning horn was misread, are covered in Explosive vs Rapid Decompression.
The autopilot flew the programmed route to Athens and entered a holding pattern over the KEA beacon. Two Hellenic Air Force F-16s intercepted the aircraft during its sixth hold. According to the AAIASB report, at 11:50 local time the left engine flamed out from fuel depletion and the aircraft began to descend. The right engine flamed out at about 7,100 feet at 12:00. At 12:03 the aircraft struck hills near Grammatiko, about 33 km north-west of Athens airport. All 121 people on board were killed.
The fuel did not run out because of a fuelling or planning error. It ran out because the aircraft kept flying for almost three hours on a trip planned for far less, with nobody able to land it.
Other fuel accidents
These two accidents do not have event pages on this site yet.
- United Airlines Flight 173, Portland, Oregon, 28 December 1978. A DC-8-61 held near Portland for about an hour while the crew dealt with a landing gear problem, and it crashed about 6 nautical miles south-east of the airport when all four engines ran out of fuel. Ten of the 189 people on board were killed. The NTSB found the captain did not properly monitor the fuel state, and that the other two crew members did not fully grasp how critical it was or did not get their concern across. The Board recommended assertiveness training for all airline crews, one of the ideas behind crew resource management (see What Changed After Tenerife).
- Avianca Flight 52, Cove Neck, New York, 25 January 1990. A Boeing 707 was held three times for a total of about 1 hour 17 minutes in bad weather approaching New York. After a missed approach at JFK, all four engines lost power and the aircraft crashed about 16 miles from the airport. Seventy-three of the 158 people on board were killed. The NTSB found the crew did not manage the fuel load and did not tell air traffic control they had a fuel emergency before the fuel ran out. It also cited the lack of standard, understandable terms for minimum and emergency fuel states.
What changed
The official reports led to changes in four areas.
- Fuel loads are cross-checked independently. The Lockwood inquiry recommended that fuel figures be checked independently, and that the job of calculating them be clearly assigned.
- One unit across a fleet. The inquiry recommended a single unit of measurement across Air Canada’s fleet and dripsticks that read directly in kilograms. That removes the conversion step where the Gimli error happened.
- Leaks are treated as leaks. After Air Transat 236, Airbus added a warning that compares the fuel on board with the flight plan. Training also put more weight on the fuel leak checklist. In the US, low-fuel alerting was written into the FAA’s ETOPS certification standards: an alert must appear while there is still enough fuel to divert safely.
- Crews speak up, and say “emergency”. United 173 and Avianca 52 both involved crews who knew the fuel was low but did not act on it, or did not make it clear to controllers. The NTSB’s findings fed into crew training on speaking up and into its call for standard fuel-state terminology.
Each of these accidents has a final report worth reading in full. How to Read an Air Accident Final Report explains how they are laid out.
The three live cases compared
| Flight | Aircraft | Date | Why the fuel ran out | After the engines stopped | Outcome |
|---|---|---|---|---|---|
| Air Canada 143 | Boeing 767-233 | 23 July 1983 | Fuel calculated with a pounds-per-litre factor instead of kilograms-per-litre; gauges inoperative | Glided to the disused Gimli air base; landed with a sideslip | All 69 survived |
| Air Transat 236 | Airbus A330-243 | 24 August 2001 | Fuel line chafed through after an engine change; crossfeed sent fuel to the leak | Second engine stopped about 65 nm from Lajes; glided to runway 33 | All 306 survived |
| Helios 522 | Boeing 737-31S | 14 August 2005 | Aircraft flew on autopilot for almost three hours after the crew were incapacitated by hypoxia | Descended after the left engine stopped; struck hills near Grammatiko | All 121 killed |
Watch in the library
- Air Canada Flight 143: Pearson’s Sideslip Landing at Gimli
- Air Transat Flight 236: Fuel Leak, Glide and Lajes Landing
- Helios Airways Flight 522: Inside the Ghost Plane Inquiry
Every video on this subject is on the Fuel exhaustion page.
Sources
- Board of Inquiry appointed by the Government of Canada: Final Report of the Board of Inquiry into Air Canada Flight 143 (Commissioner George H. Lockwood), April 1985
- Federal Aviation Administration: Lessons Learned: Air Transat Flight 236, Airbus A330-243, C-GITS, Lajes, Azores (summary of the GPIAA final report)
- Air Accident Investigation and Aviation Safety Board (AAIASB), Greece: Aircraft Accident Report 11/2006, Helios Airways Flight HCY522, Boeing 737-31S, Grammatiko
- National Transportation Safety Board: Aircraft Accident Report NTSB-AAR-79-7, United Airlines Flight 173, DC-8-61, Portland, Oregon, December 28, 1978
- National Transportation Safety Board: Aircraft Accident Report NTSB/AAR-91/04, Avianca Flight 52, Boeing 707-321B, Fuel Exhaustion, Cove Neck, New York, January 25, 1990
- Australian Transport Safety Bureau: Avoidable Accidents No. 5: Starved and exhausted: Fuel management aviation accidents (AR-2011-112, 2013)
Last checked 31 October 2026.