The Stabilized Approach: The One Habit That Fixes Most Bad Landings

9 April 2026 ·Admin ·6 min read

Last week we looked at what makes a landing smooth and the five errors that ruin touchdowns. Four of those five have something in common: by the time they hurt you, they are old news. The float from excess speed, the drop from a mistimed flare, the scramble to fix drift in the last seconds - almost all of it was set up hundreds of feet earlier, on an approach that was never really under control.

Airlines worked this out decades ago, wrote it into a single rule, and in doing so quietly fixed most bad landings across an entire industry. The rule is the stabilized approach, and it is the highest-value habit a sim pilot can steal from the professionals.

What airlines learned the hard way

Through the 1990s, flight safety researchers kept finding the same thing in approach-and-landing accident data: the landing rarely failed on its own. Runway overruns, hard landings and loss-of-control events overwhelmingly began as approaches that were too high, too fast, not configured in time - and continued to a landing anyway. The Flight Safety Foundation's landmark studies of that era found unstabilized approaches were a factor in a large share of approach-and-landing accidents, and that the decision to continue an unstable approach - rather than any lack of stick-and-rudder skill - was the recurring failure.

The industry's answer was not better flaring technique. It was a gate: a height on final at which the approach must meet a short list of criteria, and a hard rule that if it doesn't, you go around. No judgement calls, no "I can save this". Stable or around.

The criteria

Wording varies between operators, but the classic stabilized approach criteria look like this. By the gate height, the aircraft is:

On the correct flight path - established on the glidepath and the extended centreline, not converging on them.
On speed - within a small band of the reference approach speed, typically VRef to VRef+10, and not decelerating through it.
In the landing configuration - gear down, landing flap set. Nothing left to move.
At a normal descent rate - usually no more than 1,000 fpm on a standard 3° path.
At a normal power setting - spooled and stable, not idle-diving to lose height and not dragging it in on high power.
Complete - checklists done, briefings done. The remaining workload is flying.

The standard airline gates are 1,000 feet above the runway in instrument conditions and 500 feet in visual conditions. Miss any criterion at the gate, or fall out of tolerance below it, and the procedure is a go-around - not because the landing is guaranteed to fail, but because the odds have moved against you and the runway will still be there in six minutes.

Why an unstable final almost guarantees a poor landing

It helps to understand why the rule works, because the mechanism is exactly what sim pilots feel on a bad day.

The first reason is energy. An approach is a controlled dissipation of energy, and the final few hundred feet have almost no capacity to absorb error. Arrive 15 knots fast at 500 feet in a light aircraft and there is simply nowhere for that energy to go before the threshold - you will float. Arrive high and correct with a dive, and the height problem becomes a speed problem three hundred feet later. Low and slow is worse still: now you are dragging the aeroplane in on power, with reduced margin above the stall and a descent rate that will build viciously if the power comes off early. None of these states is unflyable. All of them arrive at the flare with a problem the flare cannot fix.

The second reason is workload. The flare is a precision task that needs your full attention - height perception, drift, timing. On a stable approach, the last 500 feet are almost restful: the aeroplane is trimmed, the power is set, the picture is constant, and you can give the touchdown everything. On an unstable one, you spend those same seconds making large corrections, and you arrive at the flare behind the aeroplane, reacting late to whatever happens next. Two pilots of identical skill will land wildly differently if one starts the flare current and the other starts it catching up.

The third reason is that instability compounds. A high approach becomes a fast one, a fast one floats, a float invites a forced touchdown, and the forced touchdown lands long and flat with the nosewheel. Cut the chain at 500 feet and everything downstream of it never happens.

Adapting the gates to sim flying

You do not need an airline SOP manual to use this. Here is a practical version for the aircraft most of us fly in the sim.

Light GA (C172, PA-28, and friends). Use a 500-foot gate for visual circuits. By 500 feet above the runway: full landing flap, trimmed, on the extended centreline, aiming point fixed in the windscreen, speed within 5 knots of your target final speed - roughly 61-65 knots in a C172 at typical weights - and descent rate around 400-700 fpm. If you are still steep, fast, or re-trimming, go around. In the circuit that costs you one lap.

Airliners (A320, 737, and friends). Fly the real gates: 1,000 feet in IMC, 500 in VMC. Landing flap and gear by the gate, managed/target approach speed captured, on the localiser and glideslope, sink rate under 1,000 fpm, engines spooled. The "spooled" one matters more than sim pilots think - jet engines take several seconds to respond from flight idle, and an idle-thrust approach at 500 feet has already spent margin you may want in the flare.

Turboprops and everything else. The pattern generalises: configured by the gate, on speed, on path, engines in their normal approach regime, nothing left to change but the flare.

Two honest adaptations for simulation. First, we fly approaches real crews would never see - the 20-mile straight-in you set up from the world map has no traffic pattern to organise you, so the gate is doing extra work; take it seriously. Second, sim pilots almost never go around, because nothing is at stake. Make the go-around cost something artificially: log it, fly the full missed procedure, rejoin properly. A go-around that takes eight minutes teaches approach discipline far better than one that takes a slew-key reset.

Make the gate a habit, not an aspiration

The stabilized approach only works as a rule. As a vague intention it does nothing, because the whole failure mode it protects against is the human conviction that this approach, though scruffy, can be saved - and in the sim it usually can, which is exactly how the habit of continuing gets trained in.

So borrow the airline discipline completely. Call the gate out loud if it helps ("five hundred, stable" - or "five hundred, going around"). Judge yourself not on whether the landing worked, but on whether the approach met the criteria. A firm touchdown off a stable approach is a data point; a greaser off an unstable one is a bad habit being rewarded.

Scoring your finals

The awkward part of self-enforcing a stability rule is honesty. Were you really on speed at 500 feet, or does it just feel that way now the flight is over? Memory is generous, and the sim keeps no minutes of the meeting.

This is a place where recording your flights changes the game. My FS Flights breaks every recorded approach into segments and scores the final approach on exactly the criteria above - how well you held the glidepath, your speed against reference speed, and your alignment with the runway, segment by segment down to the threshold, along with your descent rate into the touchdown. Fly a week of approaches and the pattern is unmissable: stable finals cluster around good touchdown scores, and the scruffy ones sit in a scatter of firm arrivals and long landings. It is the airline safety data, reproduced in miniature, in your own logbook - and it makes the case for the 500-foot gate more persuasively than any article can.

Next time you are on final and something is not right - a bit fast, a bit high, still trimming - remember that the landing is already being decided. Give yourself the gate, and give yourself permission to use it. The runway will still be there in six minutes.