CPAP 95th-Percentile vs Median Pressure: What the Numbers Mean

Your CPAP machine doesn't blow one steady pressure all night — it records a whole range of pressures and then summarizes them with a couple of key numbers. Two of the most useful are the median (P50) and the 95th-percentile (P95) pressure. Understanding what each one means, and especially the gap between them, tells you far more about how your therapy is going than the single "pressure" number on your app's home screen.

This is one of the most misread sections of any CPAP report, so let's break it down in plain English and connect it back to the bigger picture of reading your CPAP data.

What 'percentile pressure' means

A percentile is just a way of ranking your night from lowest pressure to highest. The P95 (95th-percentile) pressure is the pressure at or below which you spent 95% of the night. In other words, your machine was at or under that number for 19 of every 20 minutes, and above it only during the remaining 5%.

The same logic applies to any percentile:

• Median (P50): the pressure you were at or below for half the night — the "typical" pressure.
• P90: at or below for 90% of the night.
• P95: at or below for 95% of the night — closer to your peak.
• Max: the single highest pressure reached, even if only for seconds.

Why not just report the maximum? Because the max can be a one-off spike — a single big leak or a brief cluster of events — that doesn't represent your real therapy. Percentiles deliberately trim off those rare extremes so you see the pressure your airway actually needed most of the time. That makes them a much more honest summary than the peak.

This only matters if your machine varies its pressure. On a fixed-pressure CPAP, the median and P95 are essentially the same number. The percentile story comes alive on auto-adjusting machines — see APAP vs CPAP vs BiPAP for how each mode behaves.

P95 vs median (P50) — the spread tells the story

Here's the part most people miss: the single most informative thing isn't P95 or median alone — it's the gap between them.

• A small gap (say, median 9.0 and P95 9.8) means your pressure stayed in a narrow band all night. Your airway needed roughly the same support throughout, and your therapy looks stable.
• A large gap (say, median 8.0 and P95 14.0) means your pressure was climbing hard at times. A wide spread between median and P95 can indicate that your auto range is set wide, or that breathing events were repeatedly pushing the pressure up in search of the level that finally held your airway open.

A wide gap isn't automatically "bad," but it's a flag worth understanding. It can come from sleeping on your back part of the night, REM-stage airway collapse, or residual events your current range is chasing. To find out which, you have to look at your pressure curve next to your event markers — not just the summary numbers. This is exactly where most apps fall short and where the reasons your AHI stays high often hide.

SomniCharts charts your full pressure distribution and overlays it against your scored events, so you can actually see whether a high P95 lines up with clusters of apneas or with leaks — the kind of detail ResMed's myAir omits entirely. (SomniCharts imports ResMed, Philips Respironics including the DreamStation 2, and Löwenstein prisma data and explains it in plain language automatically.)

ResMed uses 95%, Philips uses 90%

This trips up anyone comparing machines or reports across brands: ResMed reports a 95th-percentile pressure, while Philips Respironics reports a 90th-percentile pressure. They're measuring the same idea — "the high end of where my pressure lived" — but at slightly different cutoffs.

The practical consequences:

• Don't compare a ResMed P95 directly to a Philips P90 as if they're identical. By definition, a P90 trims off a little less of the top end, so for the same therapy a Philips "90% pressure" will tend to read slightly lower than the ResMed-style P95 would for the same night.
• The difference is usually small, but it's real — and it matters most when you're switching CPAP brands and want to keep your data history comparable.
• Always note which percentile a report uses before drawing conclusions. A good multi-vendor tool labels this clearly instead of mashing brands together.

If you're new to the brand-specific quirks, the ResMed AirSense 11 data guide and DreamStation 1 data guide walk through what each machine records on its SD card.

What P95 reveals on APAP (hitting the ceiling vs comfortable range)

On an auto-adjusting machine (APAP), you set a range — a minimum and a maximum pressure — and the machine moves freely inside it in response to your breathing. Here, P95 is a powerful diagnostic.

Sitting comfortably inside the range. If your range is 6–15 and your P95 lands around 11, your machine has plenty of headroom. It found the pressure it needed and never ran out of room. That's the picture you want.

Hitting the ceiling. If your range is 6–12 and your P95 is sitting at 11.9 or pinned right at 12, that's a warning sign. Your machine may have wanted more pressure to clear events but couldn't go higher because it hit the cap you (or your clinician) set. A P95 jammed against the maximum, especially alongside a higher-than-expected AHI, is a classic signal that the range deserves a second look.

A floor set too high. The opposite also shows up in the data: if your minimum is high and your median sits right at that floor, you may be getting more pressure than you need for most of the night, which can drive comfort problems like aerophagia (air swallowing and bloating).

A few cautions before you read too much into one report:

• One night is noise; trends over weeks are the signal. A single night's P95 can be thrown off by alcohol, illness, congestion, or sleeping position. Look at the pattern across two to four weeks before concluding anything. (See why your AHI changes night to night.)
• A high P95 driven by leaks is misleading. Big mask leaks can confuse event detection and make the machine react oddly. Check your leak rate before blaming the pressure — large leaks can under-report your true AHI.
• Rising pressure won't fix everything. If your reports show climbing central (clear-airway) events, more pressure is not the answer — see the clinician note below.

For a deeper read on whether your numbers point to too much or too little support, the guide on signs your CPAP pressure is too high or too low pairs well with this section.

Using P95 to set a fixed pressure (clinician territory)

Here's a fact that surprises many people: P95 is the figure clinicians often use to choose a fixed CPAP pressure. The logic is elegant — if a single steady pressure equal to your P95 would have been enough to handle your airway 95% of the night on auto, it's a reasonable candidate for a comfortable, effective fixed setting.

That's the concept. The execution belongs to your provider, and here's why this section is firmly clinician territory rather than a DIY checklist:

• The treatment goal for adults on CPAP is generally a residual AHI below 5 events per hour (the AASM defines an "optimal" titration as reducing AHI under 5). Some clinicians aim lower for fuller symptom control when it's comfortably achievable, but there's no formal guideline setting "below 1 or 2" as a universal target, and goals are individualized with your provider.
• A fixed pressure trades the machine's flexibility for simplicity, which can help some people (less pressure swing, less aerophagia) and hurt others (no headroom for a bad night). That's a clinical judgment call.
• Your P95 itself depends on how your auto range was configured. A P95 measured under a poorly set range isn't a clean target — it has to be interpreted, not copied.

This is the heart of the YMYL line on CPAP self-management: use your data to have an informed conversation, not to change your own prescription. Reading the numbers is empowering; quietly adjusting your prescribed settings to chase them is not. The guide on whether you can adjust your own CPAP pressure covers exactly what the data can and can't tell you here.

A word on central apneas and pressure

If your reports show a rise in central (clear-airway) events as pressure climbs, do not raise pressure on your own to chase them. CPAP is designed to splint the airway open and reliably treats obstructive apneas, but it does not directly correct the unstable breathing drive behind central events. A climb in central events on CPAP can reflect treatment-emergent central sleep apnea (TECSA), which appears in roughly 5–15% of PAP titrations and often resolves on its own within weeks to a few months of continued CPAP (reported spontaneous resolution around 60–80%). Higher pressure won't fix central apneas and can sometimes provoke them. Persistent central events — especially with returning symptoms, heavy daytime sleepiness, deep oxygen drops, or new medications or heart conditions — warrant clinician evaluation. The full story is in treatment-emergent central sleep apnea on CPAP.

Bringing your pressure data to your provider

Your median and P95 are some of the most useful numbers you can put in front of your sleep clinician, because they summarize behavior over time rather than a single moment. To make the conversation productive:

1. Bring trends, not a screenshot. Pull together two to four weeks of nights so your provider sees the pattern, not one outlier.
2. Note your range alongside your P95. "My range is 6–12 and my P95 is pinned at 11.9 most nights" is a precise, actionable observation.
3. Pair pressure with events and leaks. A high P95 means one thing with low leak and clean events, and something very different when leaks are high — context changes the interpretation entirely.
4. Ask questions, don't prescribe answers. "Could my upper limit be too low?" invites the right discussion; deciding the answer yourself doesn't.

If you'd like everything in one organized place, a CPAP compliance report bundles usage, AHI, leaks, and pressure for your doctor or insurer. And because so much of this hinges on seeing pressure, events, and leaks together, a tool that charts your full distribution in plain language — across ResMed, Philips, and Löwenstein machines — turns a cryptic summary screen into a conversation you can actually lead. That's the gap SomniCharts is built to close, where brand apps like myAir and DreamMapper (which shut down in January 2026) leave you with a score and little else.

The takeaway: don't fixate on a single pressure number. Watch your median, watch your P95, and pay closest attention to the gap between them and to your P95 relative to your range — over weeks, not nights. That's what the percentiles are really trying to tell you.