CPAP Leak Rate: What's Acceptable and How to Fix High Leaks

Your leak rate is one of the most misunderstood numbers on a CPAP report — and one of the most important. A leak that climbs too high doesn't just dry out your mouth or wake you up; it can quietly make your therapy look better than it is by hiding the very events your machine is supposed to catch. This guide explains what counts as an acceptable leak, why ResMed and Philips put different numbers on your screen, and what to do when the leak rate is too high.

This article is part of our Reading Your CPAP Data pillar. If you're new to interpreting your nightly summary, start with our plain-English overview of AHI, leak rate, and pressure and come back here.

Intentional (vent) leak vs unintentional leak

Every modern CPAP mask is designed to leak. That sounds wrong, but it's essential: your mask has small exhalation vents that continuously flush out the carbon dioxide you breathe out, so you don't rebreathe it. This is called intentional leak or vent leak, and it rises and falls with your pressure — higher pressure pushes more air through the same vent holes.

The number that matters for therapy quality is the unintentional leak: air escaping where it shouldn't — around the edge of the cushion, out of your mouth, or through a poorly seated seal. This is the air your machine can't account for, and too much of it undermines the pressure reaching your airway.

So when you read your report, the key question is: which leak is this number measuring? That's where the two big manufacturers diverge.

The numbers that matter — ResMed 24 L/min excess leak at the 95th percentile

ResMed machines (AirSense, AirCurve) report excess leak — the unintentional portion only, with the expected vent leak already subtracted. ResMed's large-leak threshold is approximately 24 L/min at the 95th percentile.

Two terms to unpack there:

• 24 L/min means 24 liters of unaccounted-for air escaping per minute.
• 95th percentile means the value that 95% of your night stayed at or below. In other words, you can have brief spikes, but if your 95th-percentile leak is creeping toward 24 L/min, leak was a sustained problem — not a one-second blip.

ResMed treats 24 L/min as the point where leak begins to compromise the machine's ability to deliver and measure therapy. Below that, occasional spikes are usually fine. At or above it, the machine may flag a Large Leak period, and your data for that stretch becomes unreliable.

A quick caution that runs through all of CPAP data: a single bad night is noise. A leak number that's high for weeks is a trend worth acting on. Watch the pattern, not one report.

Why ResMed reports excess leak and Philips reports total leak

Here's the part that confuses anyone comparing two machines: ResMed and Philips measure leak from different baselines.

• ResMed subtracts the expected vent leak and shows you only the excess (unintentional) leak. A "good" ResMed leak number can read quite low — often single digits.
• Philips Respironics (DreamStation, System One) reports total leak — the unintentional leak plus the intentional vent leak baked in. That makes Philips leak numbers look much larger, because they include all the air your mask is supposed to vent.

This is why you can't compare a ResMed leak figure to a Philips leak figure directly. A Philips total leak of 40 L/min might be perfectly healthy (mostly vent leak), while a ResMed excess leak of 40 L/min would be a serious problem. The numbers measure different things.

This cross-machine confusion is exactly the kind of thing SomniCharts is built to clear up. It imports ResMed, Philips Respironics (including the encrypted DreamStation 2), and Löwenstein prisma data, normalizes how leak is presented, and flags large-leak nights in plain language — so you're not left translating between two different yardsticks. If you're switching brands, see keeping your data history across vendors.

How a large leak invalidates your AHI

This is the trust-building point, and it's the reason leak deserves your attention even more than AHI on a bad night.

Your machine detects apneas and hypopneas by sensing changes in airflow and pressure. When a large leak is dumping air out of the system, the machine can't reliably read what's happening at your airway. During flagged large-leak periods, many machines stop scoring events — or score them inaccurately — because the airflow signal is contaminated.

The result is a dangerous illusion: a large leak can falsely lower your AHI. You might open your report, see an AHI of 1.2, and feel great — when in reality the machine simply couldn't see the events happening behind a wall of escaping air.

So the rule is:

A low AHI on a high-leak night is not a victory. It's a question mark.

Always read leak and AHI together. If your AHI dropped and your leak spiked on the same night, suspect a leak artifact before you celebrate. For more on what drives a genuinely high AHI, see why your AHI is high on CPAP, and for what a healthy residual number looks like, what is a good AHI. For context, the widely used benchmark for effective therapy is a residual AHI below 5 events per hour — but that benchmark only means something when leak is controlled enough for the machine to measure accurately.

This is where SomniCharts' analysis layer earns its keep: it shows you when leaks line up with event spikes (or suspicious event drops), so you can tell whether a low AHI is real or a leak artifact — without manually cross-referencing two charts.

Common causes — seal, mouth leak, worn cushion

Most high leak traces back to one of a handful of causes. Work through them in roughly this order:

1. Mask fit and seal. The most common culprit. A mask that's too big, too small, or the wrong style for your face will leak no matter how tight you crank it. (Overtightening usually makes leaks worse by deforming the cushion.) Our mask types and fit guide walks through nasal, pillow, and full-face trade-offs.
2. Mouth leak. If you use a nasal or pillow mask and your mouth falls open during sleep, pressurized air escapes out through your lips. This is a leak type, not a mask-fit failure, and it often shows up as leak spikes later in the night during deeper sleep.
3. Worn or dirty cushion. Silicone cushions degrade. After a few months they lose their springiness and stop sealing — a slow, creeping rise in leak over weeks is a classic sign. Regular cleaning extends cushion life; see how to clean your CPAP.
4. Position and movement. Side sleepers can break the seal against the pillow; tube tug from a poorly routed hose pulls the mask off-center.

A frequently missed connection: dry mouth is usually a downstream symptom of leak, not a separate problem. If you wake up parched, a mouth leak is often pulling dry air across your tongue all night. Treating the leak frequently fixes the dry mouth. We cover both sides in CPAP dry mouth and mouth breathing and the data-driven mask leak and mouth leak fixes.

When leaks get severe enough that the mask effectively comes off, your report shows them as Large-Leak or Mask-Off events — gaps in your data. We explain those report markers in mask off and large-leak events.

Reading leak as a timeline, not a single average

The biggest mistake people make with leak is treating it as one number. Two nights can show the exact same average leak for completely different reasons:

• Steady, low-level leak all night — often a marginal seal or slightly wrong mask size.
• Long stretches of zero leak with sharp spikes — often mouth leak in deep sleep, or repositioning at predictable times.

These need different fixes, and a single average hides the difference entirely. That's why the leak graph over time matters more than the summary statistic. Look at when the spikes happen:

• Spikes clustered in the early morning hours point to mouth leak as you enter deeper sleep.
• Spikes that line up with REM periods suggest movement or position changes.
• A leak floor that has slowly risen week over week points to a worn cushion that needs replacing.

This timeline view is also how you confirm whether a low AHI is trustworthy: overlay the leak graph on the event markers and check whether your "good" stretches were actually low-leak stretches. SomniCharts builds that overlay for you automatically and flags large-leak nights across all three supported vendors — turning a wall of numbers into a clear answer about whether last night's data can be believed.

Frequently asked questions

What is an acceptable CPAP leak rate on a ResMed machine? On ResMed devices, you want your 95th-percentile excess leak under 24 L/min. ResMed already subtracts the expected vent leak, so a good number is often in the single digits. Brief spikes are normal; a sustained 95th-percentile reading near or above 24 L/min is the signal to investigate your seal.

Why is my Philips leak number so much higher than a ResMed's? Because Philips reports total leak (intentional vent leak + unintentional leak), while ResMed reports only the excess unintentional portion. A higher Philips number is largely expected vent air and isn't directly comparable to a ResMed figure.

Can a leak make my AHI look better than it really is? Yes. During large-leak periods the machine can't reliably detect events, so it may under-score or stop scoring entirely — falsely lowering your AHI. Always read leak and AHI together; a low AHI on a high-leak night should be treated with suspicion.

Does high leak cause dry mouth? Often, yes. Mouth leak in particular pulls dry, pressurized air across your mouth all night. Dry mouth is frequently a symptom of leak rather than a separate issue, and fixing the leak usually resolves it.

If your leak numbers stay high after working through fit, cushion, and mouth-leak fixes, use your data to start an informed conversation with your provider or DME — a different mask style or added chin support is sometimes the real answer.