I was a passenger on a Southwest Airlines flight to Chicago Midway about a month before this mishap and noticed the crew elected to use idle reverse on a runway I would consider fairly short for a Boeing 737. A month after the mishap I was on another Southwest flight flying into a longer runway when the pilots elected to use full reverse. The company apparently had a philosophy change. For the better. After this accident report was released, I found out that Southwest Airlines policy during the time of the accident was for the reversers to be used immediately on all landings. I guess not all Southwest pilots got the word. As I write this, over ten years later, I still notice not all their pilots use reverse thrust for every landing.

— James Albright





NTSB Report, Figure 1.

The classic debate when I was at Andrews was to use full or idle reverse. "Why ruin an otherwise good landing with all the noise and vibration from the reversers?" My view: the reversers are only effective at high speeds; you should use as much as you can as soon as you can until it is obvious you can stop without them. More on this: Smoothness.

Prompt application of the reversers might have kept this airplane on the runway, we aren't sure because there were other complications. But the pilots were also set up, in a way. They were responsible for flying other types of 737s in the Southwest fleet. Their performance computers did not take credit for the impact of thrust reversers during landing on the -300 and -500 models these pilots flew. But on the -700 model, like this one, the performance computer did take credit. It could very well be that these pilots assumed their landing performance was going to be better than predicted by the computer.

1 — Accident report

2 — Narrative

3 — Analysis

4 — Cause



Accident report

  • Date: 08 DEC 2005
  • Time: 19:14
  • Type: Boeing 737-7H4
  • Operator: Southwest Airlines
  • Registration: N471WN
  • Fatalities: 0 of 5 crew, 0 of 103 passengers, 1 on the ground
  • Aircraft Fate: Repaired
  • Phase: Landing
  • Airports: (Departure) Baltimore/Washington International Airport, MD (BWI) (BWI/KBWI), United States of America; (Destination) Chicago-Midway Airport, IL (MDW) (MDW/KMDW), United States of America



  • On December 8, 2005, about 1914 central standard time, Southwest Airlines (SWA) flight 1248, a Boeing 737-7H4, N471WN, ran off the departure end of runway 31 center (31C) after landing at Chicago Midway International Airport (MDW), Chicago, Illinois. The airplane rolled through a blast fence, an airport perimeter fence, and onto an adjacent roadway, where it struck an automobile before coming to a stop. A child in the automobile was killed, one automobile occupant received serious injuries, and three other automobile occupants received minor injuries. Eighteen of the 103 airplane occupants (98 passengers, 3 flight attendants, and 2 pilots) received minor injuries, and the airplane was substantially damaged. The airplane was being operated under the provisions of 14 Code of Federal Regulations (CFR) Part 121 and had departed from Baltimore/Washington International Thurgood Marshall Airport (BWI), Baltimore, Maryland, about 1758 eastern standard time. Instrument meteorological conditions prevailed at the time of the accident flight, which operated on an instrument flight rules flight plan.
  • The pilots reported that they had thoroughly reviewed the two weather information and dispatch documents they received from dispatch before they left BWI. A third document authorizing the release of the accident flight was prepared but was not delivered to the pilots before departure. This document revised the expected landing winds (from “calm” to “090° at 11 knots”), runway braking action (from “wet-good” to “wet-fair”), and landing runway (from 04R to 31C) based on the changing weather. The pilots stated that they subsequently received updated MDW weather information and runway condition/braking action reports for runway 31C, which was the runway in use at MDW at the time. Post-accident interviews with the pilots and evidence from cockpit voice recorder (CVR) data and air traffic control (ATC) communications indicated that the runway 31C braking action reports were mixed, reporting good or fair braking action for the first half of the runway and poor braking action for the second half.
  • The pilots also stated (and CVR evidence confirmed) that they reviewed and discussed the company’s new autobrake system procedures while en route from BWI to MDW; the accident landing was the first time either pilot landed using autobrakes.
  • About 1833:17, as the airplane was nearing MDW at an assigned altitude of 10,000 feet, ATC issued the pilots instructions to enter a holding pattern. (ATC indicated that the hold was because of runway-clearing snowplow operations at MDW.) About 1844:04, the pilots advised ATC that they were entering the holding pattern at 10,000 feet. The first officer stated that, while in the holding pattern, he entered the updated weather and runway conditions and wind information (090° at 11 knots) in the on board performance computer (OPC) to determine the landing distance required for runway 31C. The reported wind conditions resulted in a computed tailwind component of 8 knots. All SWA 737s are limited to landing with a 10-knot or less tailwind component under all runway surface conditions. Additionally, SWA policies and flight operations manuals indicate that the company does not authorize landings on runways with more than a 5‑knot tailwind component with poor braking action. Post-accident statements and CVR evidence indicated that the accident pilots were aware of these limitations and believed that they would be unable to land at MDW if the braking action was reported poor for the full length of the runway.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

Their manual also prohibited landing with more than a 5 knot tailwind when the condition was poor and also stipulated that when the runway had multiple conditions they were to assume the worst condition for the entire runway. The crew was aware of the 5 knot limit but not of the need to consider the entire runway's braking action as poor when part of it was judged fair.

The first officer entered multiple scenarios into the OPC, entering fair and poor pilot braking action reports separately because the OPC was not designed to accept mixed braking action report inputs. Based on the first officer’s inputs, the OPC estimated that the airplane would stop about 560 feet before the departure end of the runway with fair braking action and about 40 feet before the departure end of the runway with poor braking action. The pilots stated that they decided that, consistent with SWA policies, they would divert to one of their alternate destinations (Kansas City or St. Louis, Missouri) if the tailwind component increased to above 10 knots or if pilot braking action reports indicated poor braking action for the full length of the runway. The automatic terminal information service (ATIS) reported a runway visual range (RVR) for runway 31C of about 5,000 feet.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

Their performance computer was designed to limit the tailwind entry to 5 knots for a poor runway condition, so the "40 feet before the departure end" prediction could be argued to be invalid.

  • About 1854:10, ATC began providing the pilots with radar vectors and descent instructions as they departed the holding pattern for the final approach course for the instrument landing system (ILS) approach to runway 31C. At that time, the RVR was reported as 4,500 feet variable to 5,000 feet, and the ATIS was reporting winds from 100° at 11 knots. About 1903:44, ATC cleared the pilots to intercept the runway 31C localizer. Less than a minute later, ATC cleared them for the approach and advised that the braking action reported for runway 31C was “fair except at the end [it’s] . . . poor.”
  • According to the CVR transcript, when the pilots contacted the MDW Air Traffic Control Tower (ATCT) at 1909:53.7, controllers advised them to “continue for [runway] 31C the winds zero nine zero at nine, brakin’ action reported good for the first half, poor for the second half.” About 1912:28, the first officer received a landing clearance from the ATCT. Flight data recorder (FDR) data indicated that the airplane was aligned on the runway centerline as it touched down at an airspeed of about 124 knots. The speed brakes deployed and brake pressure increased within about 1 second. Both pilots described the touchdown as “firm.”

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

All indications are that the airplane was landed where it needed to be at the correct speed.

The captain stated that he tried to deploy the thrust reversers immediately after touchdown but had difficulty moving the thrust reverser levers to the reverse thrust position. He further stated that he felt the antiskid system cycle after the airplane touched down but then felt it stop cycling and that the airplane seemed to accelerate. He said that he subsequently applied the wheel brakes manually but made no further effort to activate the thrust reversers. He told investigators that he believed that the use of the autobrake system distracted his attention from the thrust reversers after his initial attempt to deploy them.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

There was no indication of problems with the thrust reverser levers, the first officer had no difficulty with them, or did the previous captains. The report seems to not attach any credibility to the captains statement that he tried to deploy the reversers.

The first officer said that, when he sensed a decrease in the airplane’s deceleration during the landing sequence, he exclaimed, “brakes, brakes, brakes,” and manually applied the brakes. He stated that he then looked at the throttle console and saw that the thrust reverser levers were still in the stowed position. The first officer moved the captain’s hand away from the thrust reverser levers and, about 15 seconds after touchdown, initiated deployment of the thrust reversers to the maximum reverse setting. FDR evidence confirmed the systems functions described by the pilots and indicated that full thrust reverser deployment occurred about 18 seconds after touchdown.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

I think it is quite natural to feel the aircraft isn't decelerating right after touchdown even though it is, because your inner ear can confuse movement forward and down when the down vector ceases. No matter the cause, the flight data recorder shows the airplane was decelerating almost immediately after touchdown.

However, the airplane ran off the departure end of runway 31C and continued through the runway safety area (RSA), a blast fence, a navigational aid antenna, across an airport road, through an airport perimeter fence, and onto an adjacent public roadway. The airplane struck a northbound automobile on that roadway before it came to rest near an intersection located on the northwest corner of the airport.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.1

  • The main landing gear touched down about 1,250 feet beyond the runway’s approach threshold. At the time, the airplane’s ground speed was about 131 knots, its airspeed was 124 knots, its heading was 316°, and its vertical acceleration reached about 1.4 Gs.
  • The ground spoilers were fully deployed, autobrakes were applied, and vertical acceleration increased to a peak value of about 1.7 Gs within about 1.2 seconds of touchdown.
  • About 10 seconds after touchdown, engine fan speed (N1) decreased from about 32 percent at touchdown to about 20 percent, where it remained for about 8 seconds.
  • Autobrakes were deactivated about 12 seconds after touchdown, and pilot‑commanded brake pressure increased to 3,000 pounds per square inch (psi).
  • The first indication of thrust reverser activity was recorded about 15 seconds after touchdown, with full deployment about 18 seconds after touchdown. N1 reached 80 percent about 9 seconds later (about 27 seconds after touchdown) at a ground speed of about 62 knots.
  • The thrust reversers were fully deployed and the brake pressure was 3,000 psi when the nose landing gear departed the runway overrun at a speed of about 53 knots.
  • The airplane came to a stop about 500 feet beyond the end of the runway on a heading of about 340° with a collapsed nose landing gear about 8 seconds after it departed the runway overrun.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.11.12



  • Simulations that replicated the accident airplane configuration, use of deceleration devices, and weather and runway conditions showed that, under these conditions, the airplane would have required about another 753 feet beyond the end of the runway to come to a stop. A simulation case performed under the same conditions but with an equivalent headwind instead of tailwind component showed that the airplane could have stopped about 584 feet before the departure end of the runway. A simulation case performed under identical airplane configuration, weather, and runway conditions, but using SWA routine/planned deceleration procedures to decelerate, showed that the airplane would likely have stopped about 1,351 feet beyond the end of the runway. Under the same conditions and if the pilots had used Boeing’s reverse thrust procedures, simulations showed that the airplane would likely have stopped about 531 feet beyond the end of the runway. However, simulations in which maximum reverse thrust was selected 2 seconds after touchdown and maintained until the airplane came to a complete stop showed that the airplane could have stopped about 271 feet before the departure end of the runway.
  • Note: SWA planned deceleration procedures specified that reverse thrust be selected within 2 seconds of touchdown and maintained until the airplane decelerated through 80 knots, followed by smooth throttle movement to forward idle thrust as the airplane decelerated from 80 to 60 knots. 36 Boeing’s published reverse thrust procedures were similar to SWA’s except that thrust reversers were to be maintained until the airplane decelerated through 60 knots, followed by smooth throttle movement to reverse idle thrust as the airplane decelerated from 60 to 30 knots.

  • The Safety Board’s study showed that timely and sustained application of reverse thrust (including maximum reverse thrust as needed) could have been used to stop the accident airplane on the runway.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.16.1

[NTSB Aircraft Accident Report, AAR-07/06, ¶1.16.1]

  • SWA pilots are trained to move the thrust reverser levers aft smoothly and promptly after touchdown and to keep their forearms on the throttle knobs to keep the throttles at idle during thrust reverser deployment. If the throttle levers are forward of the idle detent by about 1/4 inch, the reverse thrust levers cannot be operated. A post-accident survey of SWA personnel and pilots, a review of SWA maintenance records, and a review of aviation safety reporting system (ASRS) data revealed no evidence of systemic thrust reverser difficulties. Several SWA pilots did report difficulties deploying the thrust reversers when they tried to move the reverse thrust levers past the interlock position too rapidly; those pilots reported that the levers moved readily when they tried to deploy the thrust reversers again after the interlocks released. Post-accident interviews with the previous 10 flight crews for the accident airplane revealed that they reported no difficulty deploying the thrust reversers.
  • According to SWA personnel, there was no policy allowing pilots to apply a credit for the use of reverse thrust during their landing distance assessments (which would increase the calculated stopping margin for a landing) until 1998, when a reverse thrust credit was incorporated into landing distance calculations for the 737-700 model only. SWA pilots received instruction regarding the reverse thrust credit during differences training when qualifying for the 737-700 model. However, until 1 week before the accident, the information in two out of three FOM locations incorrectly indicated that reverse thrust was not included in 737-700 OPC landing distance calculations. At the time of the accident, the information was correct in two of the three FOM locations. Most SWA pilots interviewed after the accident were aware of the OPC reverse thrust credit for the 737-700; however, some were not.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.17.2

The OPC is a lap top computer "On Board Performance Computer."

  • At the time of the accident, SWA planned to implement a policy requiring the use of autobrakes under certain landing conditions on December 12, 2005 (4 days after the accident). Previously, SWA’s policy did not permit the use of autobrakes because the company’s fleet was not then fully equipped with autobrakes. As the fleet became fully equipped, the company took steps to implement autobrake use. In preparation, SWA provided its pilots with a self-study training module on the autobrake system and related SWA procedures (which both accident pilots had completed). SWA pilots were also provided with a series of bulletins regarding the repeatedly delayed start-date for autobrake usage. The most recent of these bulletins was issued December 8, 2005, and noted that the company’s autobrake procedures and policies were to be used by SWA pilots beginning December 12, 2005, and therefore were not in effect (or authorized) on the day of the accident.
  • During post-accident interviews, the pilots told Safety Board investigators that they had read the daily read-before-flight (RBF) letters before the accident flight but that they failed to notice the new delay in autobrake procedure implementation. CVR evidence and post-accident statements indicated that they both believed that the autobrake policy was in effect for the accident flight. A previous autobrake-related RBF letter indicated that the autobrake policy would be in effect as soon as materials were available in the cockpit. On the day of the accident, “flow” cards and checklists with information regarding autobrake procedures had been placed in SWA airplanes.
  • When using autobrakes, pilots can select from several autobrake system settings, including the following:
    • Maximum (MAX): Should be used when minimum stopping distance is required.
    • Medium (MED): Should be used for wet or slippery runways or when landing rollout distance is limited.
    • Minimum (MIN): These settings provide a moderate deceleration effect suitable for all routine operations.
  • SWA pilots were type-rated in all three 737 models owned by SWA (-300/-500/-700) and switched between these models on a day-by-day or flight-by-flight basis. Pilots were taught that stopping margins assumed reverse thrust credit (85 percent to MAX) for the -700 model only and not for the other two models (-300/-500). This method of calculation resulted in a more favorable stopping margin for the -700. As noted, until 1 week before the accident, OPC-related information in the SWA FOM stated that reverse thrust was not included in the landing distance calculations in two of three locations; the incorrect information remained in one location when the accident occurred but has since been corrected.
  • Post-accident examination of the data stored on the accident OPC confirmed that the pilots had entered the expected airplane touchdown weight and updated weather data. As previously noted, evidence indicated that the pilots selected WET-FAIR and WET-POOR as possible runway conditions and that the OPC estimated 560 and 40 feet, respectively, of runway remaining under those conditions. (As previously noted, SWA procedures permit company pilots to land with any positive calculated stopping margin.) Based on its assumptions, the OPC display would reflect the landing distance associated with a maximum tailwind component of 5 knots for poor braking action even if the computed tailwind component exceeded 5 knots.
  • The Safety Board’s investigation revealed that if SWA OPCs had used the actual tailwind component of 8 knots instead of the company limit of 5 knots, the stopping margin for poor braking action would have been -260 feet. Because of its negative value, this number would have been presented as bracketed white digits inside of a red block (instead of the standard black digits against white background) to alert the pilots that they could not safely land on the runway. The Board notes that calculations performed using Boeing’s more conservative data, an 8‑knot tailwind component, and poor braking action indicated that the airplane would have stopped 2,070 feet beyond the end of the runway. Similar calculations performed using fair braking action indicated that the airplane would have stopped 260 feet beyond the end of the runway. A review of SWA guidance and training regarding OPC assumptions revealed one reference in the OPC section of the flight reference manual (FRM) but none in the FOM. Further, no references were made to this topic during initial, recurrent, or OPC-related ground training.
  • SWA guidance indicates that mixed braking action reports are not unexpected during routine operations and company policy requires pilots to defer to the more critical braking action assessment when mixed braking action conditions are reported. Specifically, SWA’s FAA-approved FOM, chapter 3, “Normal Operations,” pages 3.23.1 through 3.23.5, states the following: Braking action reports less than good are classified according to the most critical term [emphasis added] (fair, poor, nil, or combinations). Operations are prohibited on all surfaces classified as nil.
  • According to SWA’s FOM, chapter 2 (“Operational Considerations”), page 2.2.6 “landing is not authorized … when wind limitations are exceeded.” The FOM, chapter 2, page 2.2.9, indicates that SWA’s maximum tailwind component for landing under poor braking action conditions is 5 knots; under all other conditions, the maximum tailwind component for landing is 10 knots.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶1.17.2

  • All of the braking action reports provided by ATC to the accident pilots were mixed and reported poor braking action on some portion of the runway. For example, 8 to 9 minutes before touchdown, the pilots received a braking action report of “fair…except at the end it’s poor,” and 3 minutes before touchdown they received a braking action report of “good for the first half, poor for the second half.” SWA policy requires pilots to defer to the more critical braking action assessment when they receive mixed braking action reports. Therefore, because “poor” braking conditions were reported for a portion of the runway and SWA guidance indicates a maximum 5-knot tailwind to land if such conditions are reported, the pilots should not have landed at MDW. The Safety Board concludes that because the pilots did not use the more critical braking action term (poor) during their landing distance assessment (which, combined with the associated tailwind limitation, would have required them to divert), they were not in compliance with SWA’s policies.
  • The pilots did not discuss interpretation of mixed runway condition reports, although their behavior and other discussion suggests that they interpreted the runway condition as closer to fair than poor. They stated during post-accident interviews (and CVR evidence indicates) that they were not aware of SWA’s guidance regarding mixed braking action reports. On the day of the accident, three preceding company aircraft (same make, model, policies, and procedures) landed with braking action reports containing the term poor and with similar wind conditions. Based on this and information obtained from post-accident interviews, it appears that other SWA pilots also were not aware of the mixed braking action report guidance in the FOM or did not adhere to it. The fact that other SWA pilots decided to land at MDW in these conditions likely influenced the accident pilots’ decision to land; the accident pilots were less likely to consider their decision to land as contrary to company guidance if other SWA crews landed in similar conditions.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶2.1

  • The accident pilots were also not aware that the 737 stopping margins computed by the SWA OPC were designed to incorporate the use of reverse thrust for the 737-700 model only, which resulted in more favorable stopping margins. Post-accident interviews with SWA pilots indicated that some (including the accident crew) assumed that none of the 737 OPC landing distance calculations took into account the use of reverse thrust. Because of this, the accident pilots believed that their intended use of reverse thrust during the landing roll would provide them with several hundred feet more stopping margin than the OPC estimated.
  • Therefore, the Safety Board concludes that if the pilots had been presented with stopping margins associated with the input winds or had known that the stopping margins calculated by the OPC for the 737-700 already assumed credit for the use of thrust reversers, the pilots may have elected to divert.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶2.2.2

Post-accident calculations showed that, if the pilots had promptly initiated and maintained maximum reverse thrust throughout the landing roll, the airplane would not have run off the end of the runway. Therefore, the Safety Board concludes that the pilots would have been able to stop the airplane on the runway if they had commanded maximum reverse thrust promptly after touchdown and maintained maximum reverse thrust to a full stop.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶2.2.2



  • The National Transportation Safety Board determines that the probable cause of this accident was the pilots’ failure to use available reverse thrust in a timely manner to safely slow or stop the airplane after landing, which resulted in a runway overrun. This failure occurred because the pilots’ first experience and lack of familiarity with the airplane’s autobrake system distracted them from thrust reverser usage during the challenging landing.
  • Contributing to the accident were Southwest Airlines’ 1) failure to provide its pilots with clear and consistent guidance and training regarding company policies and procedures related to arrival landing distance calculations; 2) programming and design of its on board performance computer, which did not present inherent assumptions in the program critical to pilot decision-making; 3) plan to implement new autobrake procedures without a familiarization period; and 4) failure to include a margin of safety in the arrival assessment to account for operational uncertainties. Also contributing to the accident was the pilots’ failure to divert to another airport given reports that included poor braking actions and a tailwind component greater than 5 knots. Contributing to the severity of the accident was the absence of an engineering materials arresting system, which was needed because of the limited runway safety area beyond the departure end of runway 31C.

Source: NTSB Aircraft Accident Report, AAR-07/06, ¶3.2


(Source material)

NTSB Aircraft Accident Report, AAR-07/06, Runway Overrun and Collision, Southwest Airlines Flight 1248, Boeing 737-7H4, N471WN, Chicago Midway International Airport, Chicago, Illinois, December 8, 2005.