Engine failure during takeoff is the single most unforgiving emergency a multiengine pilot will brief, and it is the emergency most often mishandled in accident reports. The reason is not a lack of knowledge about which engine failed or which rudder to push. The reason is timing. The correct action changes, sometimes in a span of one or two seconds, depending on where the airplane is in relation to a specific airspeed: minimum controllable airspeed, or Vmc. FAA-H-8083-3C, Chapter 13, builds the entire takeoff emergency framework around this speed, and 14 CFR 23.149 defines how Vmc is established for the airplane in the first place. Understanding the relationship between Vmc, liftoff, and climb performance is what separates a survivable engine failure from a fatal one.
This article lays out the decision framework exactly as the Airplane Flying Handbook presents it: what to do before Vmc, what to do after Vmc, why the dividing line is not negotiable, and why the temptation to "just fly it off" below Vmc has killed pilots who had otherwise flawless technique.
Vmc: The Line the Whole Decision Is Built On
Vmc, minimum control speed, is the calibrated airspeed at which, when the critical engine is suddenly made inoperative, it is possible to maintain control of the airplane with that engine still windmilling, up to 5 degrees of bank toward the operative engine, and to maintain straight flight with not more than a 5-degree bank. It is established under 14 CFR 23.149 using a specific, standardized set of conditions: maximum available takeoff power on the operating engine, rearmost center of gravity, flaps in the takeoff position, gear retracted, and the airplane trimmed for takeoff. The published Vmc, marked with a red radial line on the airspeed indicator, is a demonstrated speed under those exact conditions.
The critical fact for takeoff decision-making is this: Vmc is an airspeed at which control is possible, not a speed at which climb is guaranteed. Below Vmc, if the critical engine fails, there is not enough rudder authority to counteract the asymmetric thrust and yawing/rolling moment of the operating engine at full power. The airplane will yaw and roll toward the dead engine, and if the pilot attempts to hold heading with rudder alone, the airplane runs out of rudder authority and departs controlled flight. This is not a performance problem to be managed by pitch and power. It is a control problem, and no amount of skill overcomes insufficient rudder authority at insufficient airspeed.
Vmc Is Not Fixed — It Moves With Conditions
Published Vmc assumes ideal conditions for control: forward CG is actually the higher-risk condition for Vmc because control authority requirements shift, but manufacturers demonstrate Vmc at the condition that produces the highest value, so operating outside those conditions can put actual Vmc above the red line. Density altitude, for example, reduces the operating engine's power output, which actually lowers the thrust-induced yawing moment and can lower demonstrated Vmc — but density altitude also degrades climb performance, which is the more dangerous factor discussed later. Conversely, a more aft CG shortens the arm the rudder has to work with, increasing Vmc. Bank angle away from the operating engine, an out-of-rigging airplane, or an inoperative critical-engine assumption error can all push actual required control speed above the number on the dial. This is why the Handbook stresses flying a stabilized takeoff briefing and respecting Vmc with margin, not treating it as a bare minimum to graze.
The Takeoff Briefing: Deciding Before You Need To Decide
Chapter 13 emphasizes that the engine-failure decision must be made and briefed before the takeoff roll begins, because there is no time to reason through it once the failure occurs. The standard briefing divides the takeoff into segments tied to speed and configuration, not arbitrary distance down the runway:
- Below Vmc, on the ground: Close the throttles, maintain directional control with rudder, brakes, and nosewheel steering, and stop on the remaining runway. This is a rejected takeoff, handled exactly as in a single-engine airplane — reduce power and stop.
- Below Vmc, airborne (should this occur): The airplane must be landed essentially straight ahead, power off if necessary, accepting whatever is off the departure end of the runway. There is no attempt to salvage the takeoff by holding the airplane in the air.
- At or above Vmc but below a safe single-engine climb speed: Directional control is possible, but climb performance may not be. The decision here depends on runway remaining, obstacles, and the specific airplane's performance — but the default posture taught in the Handbook is still to reject if runway remains, because marginal climb performance with an engine out at low altitude offers little margin for error.
- Above Vmc, positive rate established, single-engine climb performance available: This is the point where the airplane is flown. Maintain directional control, establish a pitch attitude for best single-engine climb speed, and begin cleanup.
This briefing is typically condensed into a short verbal call before every takeoff: an airspeed to identify as the decision point, and a plan for what happens below versus above it. The specific numbers vary by airplane and are found in the AFM/POH, but the logic is universal and traces directly to the Vmc/Vyse relationship.
Why You Cannot "Fly Through" an Engine Failure Below Vmc
The single most dangerous instinct in a below-Vmc engine failure is to try to keep the airplane flying — to trade whatever airspeed exists for altitude, assuming that climbing away is inherently safer than staying on the ground. It is not. Below Vmc, full rudder deflection is insufficient to stop the yaw toward the dead engine. As the nose yaws, the airplane begins to roll in the same direction due to the sideslip that develops, a self-reinforcing departure from controlled flight generally referred to as loss of directional control. Once this begins at low altitude, there is rarely enough altitude or airspeed to recover. The airplane frequently rolls past 90 degrees of bank and strikes the ground in a nearly vertical attitude. This accident signature is well documented and is precisely why the Handbook and 23.149 tie the entire takeoff decision to Vmc rather than to liftoff itself.
The corollary is equally important: an airplane that has become airborne prematurely, below Vmc — for example, hauled off the runway by a pilot rotating early or by a gust — is not yet in a "flying" regime as far as engine-failure control is concerned. If the critical engine fails at that moment, the correct response is still to close both throttles and land, even if that means a firm touchdown or an overrun into rough terrain beyond the runway. A controlled arrival at low speed, wings level, is survivable. A loss of control below Vmc, almost never is.
Accelerate-Stop vs. Accelerate-Go
Many turbine and higher-performance twins publish formal accelerate-stop distance and accelerate-go distance data, and some light twins provide similar guidance or at least the performance charts to calculate it. The concept is straightforward:
- Accelerate-stop distance is the runway required to accelerate to a decision speed, experience an engine failure, and bring the airplane to a full stop.
- Accelerate-go distance is the runway (plus climb path) required to accelerate to that same decision speed, lose an engine, and continue the takeoff to a specified height, typically 50 feet, while climbing on the remaining engine.
Where both distances are known, the runway length available determines whether a rejected takeoff is even survivable within the remaining pavement at a given speed, and it directly informs the accelerate-stop and accelerate-go segments of the takeoff briefing. Most piston light twins do not have this performance certified or published to the same standard as transport-category airplanes, and many cannot meet a positive single-engine climb gradient at all under certain weight, altitude, and temperature combinations. This is a critical planning fact: the pilot must know, before takeoff, whether the airplane is capable of climbing on one engine under the conditions of the day. If it is not, the entire premise of "clean up and fly" above Vmc collapses, and the safest option — even above Vmc, even airborne — may still be to close the throttles and land, because a light twin flying at or below single-engine ceiling with the gear coming up is often descending, not climbing, regardless of pilot technique.
Above Vmc With Performance: Blue Line and the Cleanup Sequence
When the airplane is above Vmc, under control, and has demonstrated or briefed single-engine climb capability, the response shifts from an abort mentality to a flying mentality. The immediate goals are to identify, verify, and feather the failed engine while accelerating toward and maintaining Vyse — the best single-engine rate-of-climb speed, marked by the blue line on the airspeed indicator.
Vyse matters because it is the speed at which the airplane, on one engine, achieves its best rate of climb, and it is also generally the speed used to determine performance in the AFM's single-engine climb charts. Flying faster or slower than blue line on one engine costs climb performance the airplane often cannot spare. Immediately after an engine failure above Vmc, the priority order taught in the Handbook is:
- Maintain control — apply rudder toward the operating engine, bank 2 to 5 degrees toward the operating engine as the AFM specifies, and keep the ball slightly deflected toward the good engine rather than centered, since centering the ball with wings level actually increases the sideslip and drag in this configuration.
- Pitch for Vyse — trade excess airspeed for the correct climb pitch attitude, or lower the nose if below Vyse to accelerate to it, accepting reduced climb or a shallow descent in ground effect if necessary to reach blue line.
- Identify and verify the failed engine — using the standard "dead foot, dead engine" method along with instrument confirmation, before touching any control.
- Clean up the airplane — retract the landing gear once a positive rate is confirmed and no landing is intended, since gear drag is costly on one engine; retract flaps as appropriate per the AFM, typically after gear and after a safe altitude/speed is assured; and feather the identified, failed engine's propeller to eliminate windmilling drag, which by itself can be the difference between a climb and a descent.
- Secure the engine — mixture, fuel, magnetos/ignition, and cowl flaps as appropriate, per the emergency checklist, once the immediate flight path is stabilized.
Sequence matters. Retracting flaps before establishing positive climb performance, or delaying feathering while fumbling for a checklist, both cost precious altitude and airspeed margin that a marginal single-engine airplane does not have. The Handbook stresses that these actions should be memorized as immediate items, with the checklist used afterward to confirm and complete the process.
The Deadly Temptation: Forcing It Off Below Vmc
Accident history repeatedly features a specific failure pattern: a pilot experiences a partial or full loss of power at a speed at or just below Vmc, close to or just after liftoff, and instead of closing the throttles and accepting a landing on remaining runway, attempts to salvage the takeoff by holding the airplane in ground effect, raising the gear, or trying to climb away. Because the airplane is below Vmc, directional control cannot be maintained at full power on the remaining engine, and the airplane rolls and yaws into the ground, often inverted or in a steep bank, with little to no survivable margin. This is sometimes called the "impossible turn" of multiengine flying, though it more often does not even involve a turn back to the runway — it is simply the airplane departing controlled flight during the climb attempt itself.
The Handbook's guidance is unambiguous on this point: an engine failure below Vmc is a ground event, not an airborne one, regardless of whether the wheels have technically left the runway. The only correct response is to reduce power on the operating engine, which eliminates the asymmetric thrust causing the control problem, and land the airplane under control. This may mean minor gear or aircraft damage from an off-runway rollout, but it preserves the pilot's ability to fly the airplane rather than have the airplane depart on its own terms.
Building the Decision Into Habit
Because there is no time to calculate during the event, the entire framework must be internalized before every takeoff: know Vmc and Vyse for the airplane and the day's conditions, know whether single-engine climb performance exists at the current weight, altitude, and temperature, and brief the specific speed and runway point at which the plan shifts from reject to fly. The dividing line is not judgment calls made in the moment — it is Vmc, verified by a positive rate of climb and confirmed single-engine climb capability. Above that line, the airplane is flown: control, pitch for blue line, identify, feather, clean up. Below that line, the airplane is stopped: throttles closed, wings level, land straight ahead. Every element of Chapter 13's takeoff emergency procedures and the certification logic behind 14 CFR 23.149 exists to keep that one decision point sharp, memorized, and non-negotiable in the seconds after an engine fails.