Field Guide · For Techs

Fuel Trims Decoded
STFT, LTFT, and what they're confessing

A fuel trim is not a reading — it's the ECM admitting it had to correct. The percentage is a symptom; the shape across load and the split between banks are the diagnosis. Read them right and you separate an air problem from a fuel problem before you unbolt anything.

Start here

Short-term and long-term are one conversation

  • STFT is the live wiggle. The ECM watching the O2/AFR right now and nudging pulse-width in real time. It should hunt around zero.
  • LTFT is the learned average. When STFT keeps leaning one way, the ECM banks that correction into its long-term cells so STFT can go back to hunting near zero. LTFT is where the fault actually lives.
  • Positive = adding fuel to fight a lean condition (unmetered air or fuel shortfall). Negative = pulling fuel to fight rich. The number is the size of the lie the ECM is covering.
  • Total correction = STFT + LTFT. If LTFT is +18% and STFT is still swinging +6%, the ECM has banked a big offset and still can't keep up — that's worse than either number alone.
Rule #1: never read a trim as a single number. Read it by load (idle vs cruise) and by bank (bank 1 vs bank 2). The pattern is the diagnosis — the calculator below turns that pattern into a next test.
InteractiveTrim diagnoser — enter what you're seeing
Live diagnoser · one bank at a time

Feed it the LTFT at two load points

Read long-term fuel trim at warm idle, then at a steady 2,000–2,500 rpm part-throttle cruise, on one bank. Add whatever STFT is doing right now. The tool reads the shape across load and points you at the system to test. Drag the sliders.

Total @ idle (STFT+LTFT)
+0%
Total @ cruise (STFT+LTFT)
+0%

Trims in range healthy

Both load points are inside the normal band. Nothing to chase here — this is what a healthy bank looks like.

Read the shapeWhat idle-vs-cruise is telling you
  • Lean @ idle onlybig + at idle, cleans up at cruise — unmetered air; the leak is a huge % of idle airflow, tiny at load
  • Lean everywhere, flatsame + at idle and cruise — constant shortfall: fuel pressure/volume, or a global MAF offset
  • Lean worse @ cruisegrows with airflow — MAF under-reporting, or injectors/pump can't keep up under load
  • Rich @ idle onlybig − at idle, normal at cruise — excess fuel at low flow: leaking injector, high rail pressure, stuck purge
  • Rich everywhere− across the board — high fuel pressure, or a MAF over-reading
Localize itBy bank, and by system

One bank vs both

First cut
  • Both banks lean the same = a shared, global cause: fuel supply, base pressure, a MAF, or false air on the common intake.
  • One bank only = that bank's territory: a leak on that runner, a lazy O2 on that side, a weak injector, or an exhaust leak upstream of that sensor faking lean.

Unmetered air (vacuum leak)

Lean @ idle
  • Air the MAF never counted: intake boot cracks, gaskets, PCV, brake booster, a leaking purge valve.
  • Confirm with a smoke test, or watch the idle trim fall as you raise rpm — the tell of a leak diluting out under load.

MAF calibration

Scales with airflow
  • A contaminated or lazy MAF mis-reports and the error grows with airflow, so trims drift as you load it.
  • Cross-check MAF g/s against expected for the displacement, and against calculated load — a split between them is the proof.

Fuel delivery

Under load
  • Weak pump, clogged filter/sock, restricted line, or a regulator holding low pressure — the shortfall shows worst when demand is highest.
  • Prove it with fuel pressure and volume under load, not a keyed-on snapshot. A pump that holds pressure static can still starve at WOT.
Where trims fool good techs

The traps

  • Open loop = no valid trims. At WOT, cold start, and in DFCO the ECM ignores the O2 and runs off a table — the trim numbers are frozen or meaningless. Only trust them in closed loop, fully warm.
  • A lying sensor drags trims with it. A biased wideband, an exhaust leak upstream of the O2, or a header leak pulls the reading and the ECM "corrects" a fault that isn't fueling.
  • Pegged is not happy. Trims clip at their authority limit — commonly around ±25%, but it varies by OEM. A trim sitting at the wall means the real error is bigger than the number can show.
  • You cleared codes and didn't relearn. Freshly reset LTFT reads near zero for a while — drive it into closed loop and let it re-adapt before you judge.
  • Chasing the STFT wiggle. STFT bounces by design. The diagnosis lives in the learned LTFT and how it moves with load.
InteractiveSelf-check — tap to reveal

Q1. LTFT is +22% at idle but drops to +4% at cruise. Air problem or fuel problem — and where?

Air. A lean condition that's huge at idle and nearly gone at cruise is unmetered air — the leak is a big fraction of idle airflow and a trivial one at load. Smoke-test the intake, PCV, brake booster, and purge. A fuel-delivery problem would stay lean (or get worse) as demand climbs, not clean up.

Q2. Both banks read +15% and climbing as you load the engine. What's your prime suspect and your proof?

Lean that grows with airflow on both banks points at the MAF under-reporting or a fuel-volume limit. Both banks = a shared cause. Prove it: compare MAF g/s to the expected value for the displacement at WOT, and watch fuel pressure/volume under load. If MAF reads low versus calculated airflow, it's the sensor; if pressure sags under demand, it's delivery.

Q3. STFT and LTFT both read 0% at wide-open throttle. Does that clear the fuel system?

No — that's open loop. At WOT the ECM commands a target AFR from a table and ignores the O2, so trims are parked and prove nothing. Judge trims only in closed loop, fully warmed up, at idle and part-throttle cruise. Zeroed trims at WOT are the system being ignored, not the system being healthy.

MethodFrom a percentage to a proven part
  1. Confirm it's closed loop and warmNo cold-start, no WOT, no DFCO. Trims only mean something when the ECM is actually listening to the O2.
  2. Read both banks at idle and at cruiseFour numbers, not one. The shape across load and the split between banks is 80% of the diagnosis.
  3. Pick a side: air or fuelLean-at-idle-only = unmetered air. Lean-that-scales-with-airflow = MAF or fuel volume. Rich = too much fuel or an over-reading MAF.
  4. Prove it under the condition it failsSmoke test for air. Fuel pressure and volume under load for fuel. MAF g/s vs expected for calibration. Never order on the trim number alone.
  5. Fix, relearn, and confirm it centersAfter the repair, drive it back into closed loop and watch LTFT walk back toward zero. That's the proof the fix took.
The unfair advantage

Trims pointing two directions at once?

Lean on one bank, fine on the other, a MAF that looks in range and a wideband you're not sure you trust — that's the moment to hand it to Kate. Give her the four trim numbers, the conditions you pulled them under, fuel pressure and MAF g/s, and she'll hold the whole picture and tell you which reading to trust and the exact next test. No guessing, no "see a professional." Just the next move.