Should I look for a vacuum leak?

Probably not, and the code itself is telling you so. An unmetered air leak admits roughly a fixed amount of air regardless of how hard the engine is working. At idle the engine takes very little air, so that fixed leak is a large proportion of the total and produces a big lean error. Under load the engine takes ten or twenty times as much air and the same leak barely registers. A lean condition that appears only at high load therefore behaves in the opposite way to a leak, and the induction system is the last place to spend time rather than the first.

My fuel pressure was fine. Doesn't that rule out the pump?

Only if it was tested under load. At idle the engine needs a fraction of the fuel it needs at full throttle, so a tired pump, a partly blocked filter or a collapsing line will all hold specification perfectly. The test that matters is watching pressure while the engine is actually working — during a hard pull, under a brake-torque load, or on a dynamometer — and looking for it to fall away as demand rises. A single static reading at idle is not evidence about this code.

Is it safe to keep driving?

It is usually safe for normal gentle driving, since the fault does not appear at idle or light cruise. What you should avoid is exactly the driving that provokes it — towing, sustained hill climbs, heavy acceleration — because a lean mixture at high load raises combustion temperatures and can bring on knock, which does real damage in a way that a lean idle never does. Get it diagnosed before the next long or loaded journey.

Why does the code name bank 1 if the fuel pump serves both banks?

Because the module reports where it measured the problem, not where it originated. If the cause is a shared component like the pump or filter, you would normally expect a matching bank 2 code to arrive as well, either at the same time or shortly after. If bank 1 is complaining alone and consistently, that shifts weight toward something specific to that bank — its injectors or its own upstream sensor — rather than the shared supply. Check whether bank 2 is clean before quoting for a pump.

OBD-II trouble code

P2191: System Too Lean at Higher Load (Bank 1)

Bank 1 runs out of fuel correction only when the engine is working hard. The load qualifier is the whole diagnosis: a fault that appears only at load is proportional to flow, which quietly clears the usual suspects.

Medium severityPowertrainFuel and Air MeteringDrivable short-term

Quick facts

System: Powertrain
Category: Fuel and Air Metering
Severity: Medium severity
Drivable: Usually safe to drive short-term
Repair cost range: $120 – $1,600
DIY difficulty: Intermediate DIY

What does P2191 mean?

The module trims fuelling continuously to hold the mixture where it wants it, and it has a limited amount of authority to do that. This code says that on bank 1, under higher load, it reached the end of its authority in the add-fuel direction. At idle and light cruise it is fine. Only when the engine is asked for real output does the correction run out.

That qualifier is not a footnote. It is a filter, and applying it properly removes most of what people normally chase on a lean code.

Think about what changes between idle and load. The quantity of air entering the engine rises by an order of magnitude, and the quantity of fuel commanded rises with it. Now consider a fault that admits a fixed amount of unmetered air — a cracked vacuum hose, a leaking brake booster, a PCV valve stuck open, a split in the intake boot. That fixed leak is an enormous fraction of the small airflow the engine takes at idle, which is why it produces a large lean error there. At high load the engine is swallowing so much air that the same leak becomes a rounding error. Fixed-quantity faults get louder at idle and quieter under load, which is the exact opposite of what this code describes.

So the fault behind a load-qualified lean code is almost never a leak. It is something proportional — something whose error grows with flow rather than staying constant. Three families cover nearly all of it. A measurement that scales wrongly: an airflow sensor whose output is accurate through the middle of its range and reads high at the top, so the module thinks more air is entering than really is and fuels for a phantom. A restriction that limits flow rate: a clogged fuel filter, a collapsing supply line, a strainer blinded with debris, all of which pass enough fuel for cruising and cannot pass enough for demand. Or a pump that simply cannot hold pressure once the injectors are open long enough to draw real volume — a worn low-pressure pump, a failing high-pressure pump on a direct-injection engine, or a supply voltage problem at the pump under electrical load.

That also tells you how to test it, and this is where most diagnoses on this code go wrong. A fuel pressure reading taken at idle answers a question nobody asked. A weak pump and a restricted line both hold specification comfortably at idle, because at idle the demand is trivial. The measurement that matters is pressure held under demand — during a hard pull, or against a load on a chassis dynamometer, or at minimum under a brake-torque load — watched live rather than glanced at. Volume matters as much as pressure: a supply that meets specification statically and drops away as soon as it has to actually flow is the classic signature here, and it is invisible to a static test.

One more point on evidence. The freeze frame is unusually valuable on this code because it records the conditions at the moment of failure. Engine load, rpm, vehicle speed and calculated airflow at the point of setting tell you where in the range the fault appears, and that in turn tells you whether you are looking at the top of a sensor's curve or at a delivery limit. It is worth reading before anything is disconnected.

Common causes

Mass airflow sensor over-reporting at the top of its range while remaining accurate at idle and cruise
Restricted fuel filter or blinded in-tank strainer that passes enough for cruise and not enough for demand
Weak fuel pump that meets pressure specification at idle and cannot hold it under load
Failing high-pressure fuel pump or worn cam follower on a direct-injection engine
Collapsing or kinked fuel supply line that restricts only when flow is high
Fuel pressure regulator bleeding off under demand
Injectors on bank 1 restricted by deposits, limiting delivery at long pulse widths
Low supply voltage at the fuel pump under electrical load, reducing output when it is needed most

Symptoms

Check engine light after hard acceleration, towing, hill climbs or motorway merging
Hesitation, stumble or a flat spot that only appears under heavy throttle
Loss of power at the top of the rev range while the engine idles and cruises normally
Long term fuel trim on bank 1 normal at idle and climbing steeply as load increases
Misfire or knock complaints appearing under load as the mixture goes lean
Symptoms that worsen with a full vehicle, a trailer or a long gradient
Fuel economy largely unchanged in gentle driving
Freeze frame showing high calculated load and high rpm at the moment the code set

Diagnostic steps

1.Read the freeze frame before disconnecting anything, and note engine load, rpm, vehicle speed and calculated airflow at the moment of failure.
2.Use the load qualifier as an eliminator: a fixed unmetered air leak is loud at idle and negligible at load, so the absence of any lean complaint at idle argues strongly against vacuum leaks, PCV and booster faults.
3.Graph short and long term fuel trim against engine load rather than reading them at idle, and find the load at which the correction starts to run away.
4.Compare measured airflow at wide open throttle against what the engine should be capable of ingesting for its displacement and rpm. A sensor reading high at the top of its range shows up here.
5.Test fuel pressure under demand, watched live, not statically at idle. A pump or restriction that holds specification at idle can collapse the moment volume is required.
6.Check fuel volume as well as pressure where the equipment allows, since a restriction shows in flow before it shows in pressure.
7.Measure supply voltage at the fuel pump under load, because a pump starved of voltage produces less than one that is not.
8.Inspect the fuel filter and, on a returnless system, consider the in-tank strainer, which is a common and easily overlooked restriction.
9.On direct injection, evaluate high-pressure pump output and the condition of its drive before condemning anything downstream.

Repair cost

$120 – $1,600

Diagnosis is $110 to $280, and it is worth insisting the testing is done under load rather than at idle — a static fuel pressure check is the single most common reason this code gets misdiagnosed. A fuel filter is $60 to $250 fitted. A low-pressure fuel pump or module is $400 to $1,100. A high-pressure pump on a direct-injection engine runs $500 to $1,600 and more on some European applications. A mass airflow sensor is $110 to $450 fitted, though cleaning is worth trying first and costs almost nothing. Injector cleaning or replacement for one bank is $150 to $900 depending on whether the engine is port or direct injected.

Estimate your repair

Run the numbers for your vehicle

Open the Repair Cost Estimator with fuel pump replacement preselected. Adjust labor rate and vehicle category to fit your situation.

Estimate your repair cost

DIY vs shop

This is an intermediate DIY job. It usually involves diagnostic steps, specialty parts, and some careful work in tight spaces. If you have the tools and a service manual or trustworthy video for your specific vehicle, it is achievable in a weekend. Otherwise, a competent independent shop will be faster.