Fuel Trims Explained: What LTFT and STFT Tell You Before You Touch a Table

Fuel trims are the most useful diagnostic in the whole datalog and the most commonly skipped. As of July 2026 the reason is straightforward: they sit in the boring part of the log — the cruise before the pull — and everybody scrolls past that looking for the exciting bit.

That is backwards. Trims are the ECU telling you, in a single number, how wrong its model of your engine already is before you have asked it to do anything difficult. Read them first and half the work of a tune is diagnosed before you open a table.

What the two numbers are

Your ECU has a target air-fuel ratio at light and moderate load, and an oxygen sensor telling it what it actually achieved. The gap between those is corrected, and that correction is what fuel trims report.

Short-term fuel trim (STFT) is the immediate correction — the reflex. It moves continuously as the sensor reports, and on a healthy engine it oscillates gently around a small value. That oscillation is normal and is the closed-loop control system doing its job.

Long-term fuel trim (LTFT) is the learned average of those short-term corrections, stored in memory so the ECU starts closer to correct next time it sees similar conditions. It moves slowly, over minutes and drives rather than seconds.

Read together: LTFT is the persistent error, STFT is what is happening right now. A car with LTFT sitting at a large positive value and STFT hovering near zero has a real, understood, compensated error. A car with both swinging wildly has something changing in real time.

The glossary entry on fuel trims covers the mechanics, and the distinction between closed-loop and open-loop operation matters here — trims exist only where the ECU is listening to the sensor.

Reading trims correctly

Three rules, and skipping any of them produces a misdiagnosis.

Read at steady state, not in transitions. Trims during acceleration, deceleration or gear changes reflect transient fueling behaviour, not the steady-state error you are trying to measure. Hold a steady cruise, let the numbers settle, and read them there.

Read at more than one load point. This is the rule that turns a number into a diagnosis. Trims at idle, at light cruise and at moderate load tell you three different things, and the pattern across those points is what identifies the cause.

Read after the ECU has learned. Long-term trims immediately after a battery disconnect or a calibration reset are meaningless — they have been erased and not yet re-learned. Drive normally for a while first.

The pattern table

This is where trims stop being a number and start being an answer:

Pattern across load	What it usually means	Where to look
Large positive at idle, shrinking as load rises	Unmetered air entering downstream of the sensor	Vacuum leaks, intake boots, PCV
Positive and roughly constant across all loads	Airflow underestimated throughout, or injectors flowing less than modelled	MAF calibration or injector data
Negative and roughly constant across all loads	Airflow overestimated, or injectors flowing more than modelled	Injector scaling, MAF transfer function
Positive only at higher load	Airflow underestimated at high flow, or fuel delivery falling off	Airflow table top end, fuel pressure, duty cycle
Swinging widely and unstably	Sensor, wiring, misfire, or a genuinely unstable condition	Sensor health, ignition, mechanical
One bank very different from the other	Bank-specific mechanical or sensor issue	That bank's sensor, injectors, leak

The first row is the one worth committing to memory. A vacuum leak admits a roughly fixed volume of unmetered air. Against the small airflow at idle, that fixed volume is a large percentage error, so the ECU adds a lot of fuel. Against the much larger airflow at load, the same volume is a small percentage, so the correction shrinks. Trims that are large at idle and small at load are the signature of a leak, not a calibration error — and no amount of table editing fixes a leak.

The last row is equally decisive. Fuel trims are usually reported per bank on a V-configuration engine, and a large split between banks localizes the problem immediately to one side of the engine. That is diagnostic information you cannot get any other way from the driver's seat.

Why this matters before you tune anything

Every calibration decision downstream depends on the ECU's estimate of how much air is entering the cylinder. That estimate comes from the airflow model — whether your calibration uses a mass-air-flow or speed-density approach — and trims are the closest thing you have to a direct measurement of how wrong it is.

If the model needs a ten-percent correction at cruise, you are about to build a fueling table, a timing table and a set of limits on top of a model that is ten percent wrong. Worse, at wide-open throttle most calibrations go open loop — they stop listening to the sensor entirely and command fuel straight from the tables. The trims that were quietly saving you at cruise are switched off exactly when the consequences of being wrong are largest.

That is the whole argument for fixing trims first. Not because trims themselves matter at full load — they do not — but because they are your only easy evidence about the accuracy of the model that full-load fueling is derived from.

The VE and MAF correction calculator converts a trim reading into a specific percentage adjustment for the affected cells, and the step-by-step VE correction workflow covers how to apply it region by region without breaking the parts that were already right.

The mistakes that make trims useless

Tuning trims to zero without asking why they were not zero. If a vacuum leak is producing plus-fifteen at idle and you edit the airflow table until the number reads zero, you have not fixed anything — you have hard-coded the leak into your calibration. When the leak is repaired, or when it worsens, your table is now wrong in a way that is much harder to find. Diagnose mechanically first; calibrate second.

Reading trims once and concluding. A single reading at one load point is a data point, not a pattern. The diagnosis lives in how the number changes across load.

Ignoring a bank split. A meaningful difference between banks is one of the strongest localizing signals available and it gets averaged away by anyone reading a single combined number.

Assuming the sensor is fine. Trims are derived from the oxygen sensor. An aging or lazy sensor produces trims that describe the sensor rather than the engine. If a trim pattern makes no physical sense, question the measurement before you question the engine.

Where trims fit in the sequence

The order that works, and the reason each step precedes the next:

1. Mechanical integrity. No leaks, healthy ignition, no misfires. Trims cannot be interpreted over a broken engine.
2. Sensor health. The measurement has to be trustworthy before the number is.
3. Trims within a few percent across the load range. Now the airflow model matches reality at part throttle.
4. Full-load fueling verified on a wideband. Where trims cannot help you.
5. Timing. Last, because timing work done over unverified fueling stacks two risks and produces uninterpretable logs.

Every step in that list exists because skipping it makes the next one unreadable. That is the real reason the sequence is not negotiable — not caution for its own sake, but that out-of-order work produces data you cannot learn anything from.

If you want a second read on what your trims and full-load logs are actually saying, the datalog analysis page covers how TuneVault reviews commanded versus delivered fueling, trims, knock and intake temperature and returns the specific cells to change — or you can just start here.

Frequently asked questions

What is the difference between short-term and long-term fuel trim?

Short-term fuel trim is the ECU's immediate, moment-to-moment correction based on what the oxygen sensor is reporting right now. Long-term fuel trim is the learned average of those corrections, stored so the ECU can start from a better guess next time. Short-term is the reflex; long-term is the memory. When you read them together, long-term tells you the persistent error and short-term tells you what is happening at this instant.

What is a normal fuel trim reading?

Broadly, combined trims within a few percent of zero at steady cruise indicate an airflow model that matches reality. Consistent double-digit corrections in either direction mean something real is wrong — either the calibration's airflow estimate or the physical engine. The exact acceptable band varies by platform, but the principle does not: trims are an error signal, and a large persistent error is worth understanding before you change anything else.

Do positive fuel trims mean the engine is running lean?

Positive trims mean the ECU is adding fuel to reach its target, which implies it was measuring lean relative to what it expected. The engine itself is not necessarily running lean at that moment — the correction may already have fixed it. What positive trims tell you is the direction and size of the error in the ECU's model, not the current state of combustion.

Can fuel trims tell me if I have a vacuum leak?

They can point strongly at one. The classic vacuum-leak signature is large positive trims at idle and low load that shrink as engine load rises, because a fixed extra volume of unmetered air matters enormously against a small airflow and hardly at all against a large one. That load-dependent pattern is the tell — a leak looks very different from a mis-scaled airflow table, which tends to persist across the load range.

Are fuel trims active at wide-open throttle?

Generally no. At full load most calibrations switch to open loop, ignoring the oxygen sensor feedback and commanding fuel directly from the tables. That is exactly why a wideband sensor matters — trims will not save you at wide-open throttle, so the tables have to be right on their own. Trims are a part-throttle diagnostic tool, and their value is telling you whether the model those tables are built on is trustworthy.

Should I fix fuel trims before adding modifications?

Yes, wherever practical. Trims describe an error in the ECU's model of your engine, and every subsequent decision — fueling at load, timing, transient behaviour — is derived from that model. Adding parts on top of an already-wrong model compounds the error and makes the resulting datalogs much harder to interpret, because you can no longer tell which problem you are looking at.

Why did my fuel trims change after I reset the ECU?

Because the learned long-term values were erased and the ECU restarted from its base assumptions. It will re-learn over a period of driving that varies by platform. This is why a trim reading taken immediately after a reset is not comparable to one taken after a week of normal driving, and why diagnosing off freshly-reset trims often sends people chasing a problem that has not had time to appear.