Your CPAP Data Explained: AHI, Leak, Usage and Your myAir Score
Your CPAP Data Explained: AHI, Leak, Usage and Your myAir Score
Your machine records every night, and most people never learn to read it. Four numbers tell you whether your therapy is working, which part of your mask is failing, and when to call your clinic. Here's what each one means.
Jump to a number.
Your machine is a witness, not a doctor.
Every modern CPAP machine quietly records what happens each night: how long you used it, how much air escaped from your mask, how many times your breathing still paused, and how often the mask came off. Most people glance at a smiley face in an app and never look further. That's a shame, because those few numbers answer the questions people worry about most — is this actually working, and if not, why not?
The trick is to treat the data as a witness rather than a verdict. It's very good at telling you what happened. It's much less good at telling you why, and it can't tell you whether you're well. A perfect score on a night you felt dreadful isn't wrong — it's just answering a different question from the one you're asking.
What the data is good at
Spotting trends. A leak figure creeping up over three weeks is one of the most reliable early signs that a cushion is wearing out or headgear is stretching — usually before you'd consciously notice anything. Usage that slips from seven hours to five tells you something is waking you. Events that rise after you change mask tell you the new one isn't sealing properly. None of these is visible in a single night. All of them are obvious over a month.
What it isn't
The event count your machine reports is an estimate made from airflow, not a sleep study. It has no idea whether you were asleep, can't see your oxygen levels, and sometimes miscounts. It's a useful indicator and a good conversation starter with your clinic — not a diagnosis, and not grounds for changing anything yourself.
And the most important number of all doesn't appear on any screen: how you feel during the day. If your data looks perfect and you're still exhausted, believe the exhaustion, and tell your clinic.
Four numbers. Here's what each one is telling you.
Most machines and apps show some version of these. The names vary slightly between manufacturers; the meaning doesn't.
| Number | What it measures | Good sign | Worth acting on |
|---|---|---|---|
| Usage hours | How long the machine ran with the mask on | Most of your sleep, every night | Steady decline, or regularly stopping partway through the night |
| Leak | Air escaping beyond what the mask vent is designed to release | Low and stable from night to night | A rising trend over weeks, or sudden jumps |
| AHI / events per hour | Breathing pauses and shallow breaths still happening on therapy | Low and consistent — your clinic sets your target | A sustained rise, or high numbers with daytime sleepiness returning |
| Mask on/off | How many times the mask came off during the night | Once or twice — usually bathroom trips | Frequent removals, especially without remembering them |
The order of that table isn't accidental. Usage comes first because nothing else matters if the mask isn't on. Leak comes second because it's the number most often affected by your equipment, and the one you can most easily fix. AHI comes third because it's the closest thing to a measure of whether therapy is working — but it's also the one most easily misread. Each gets its own section below.
Usage hours: the number that matters most.
It sounds too obvious to need explaining, but usage is where therapy succeeds or fails. CPAP only works while you're wearing it. Sleep apnoea doesn't wait politely until you've had a few hours' rest — if you take the mask off at 3am, your breathing goes back to its untreated pattern for the rest of the night, and the hours you sleep without it are untreated hours.
That's why ResMed's own myAir score gives usage more weight than everything else combined, as you'll see further down.
Where the "four hours" figure comes from
You'll often read that four hours a night is enough. That figure comes from American health insurance rules, which in many cases require users to log at least four hours on 70% of nights during an early trial period before the insurer will keep paying for the machine. It's an administrative threshold for billing, not a clinical target, and it has never applied in the UK.
The clinical reality is simpler: more is better, and the aim is the whole night. Many of the benefits people notice — better concentration, steadier mood, less daytime sleepiness — build with consistent, full-night use. Your clinic may give you a personal target; if they haven't, "all of my sleep, every night" is the right one.
Reading your usage honestly
Look at two things rather than one. The first is the average — how many hours a night across a week or a month. The second is the pattern — whether there are missed nights, and whether your hours are short because you're going to bed late or because you're taking the mask off early.
Those two causes look identical in a single number and need completely different fixes. A short night because you went to bed at 1am is a sleep habit. A short night because you woke at 4am and couldn't bear the mask any longer is an equipment or comfort problem — and that one is almost always solvable.
Naps count
If you nap during the day, use the machine for that too. Apnoea doesn't distinguish between night-time sleep and an afternoon doze, and the hours appear in your record either way.
Leak: the number your equipment controls.
If you only learn to read one figure properly, make it this one. It's the earliest warning that something on your mask is wearing out.
Why the leak number can look worse than your night felt
Many machines don't report your average leak on the main screen. ResMed machines, for example, typically report a 95th percentile figure — the level that your leak stayed at or below for 95% of the night. That means a short burst of heavy leak, such as five minutes after you rolled over and dislodged the mask, can pull the headline figure up even if the rest of the night was fine.
That's useful to know for two reasons. It means one bad night is rarely worth worrying about. And it means a figure that stays high night after night is significant, because it's telling you the leak isn't a moment — it's most of the night.
What counts as "too much"?
Apps like myAir avoid raw numbers and show you a simple rating for mask seal, which is the easiest place to start. If you look at detailed reports from a ResMed machine, you'll often see around 24 litres per minute quoted by experienced users as the point at which leak is considered large. That figure circulates widely on patient forums rather than being something we've been able to find in ResMed's own consumer guidance, so treat it as a rough marker rather than a rule — and ask your clinic what they'd want to see for your setup.
Honestly, the trend matters more than any threshold. A leak figure that was consistently low for months and has crept up over the last few weeks is telling you something, whatever the exact number.
The five usual causes of rising leak
A worn cushion. The fastest-wearing part of any mask, and the commonest cause by a distance. Silicone softens and loses its shape with facial oils and washing. If your leak has risen gradually and your cushion is more than a couple of months old, start here.
Stretched headgear. Elastic loses its recovery invisibly. If you've fitted a new cushion and the leak is still high, the straps are usually why.
Over-tightening. Counter-intuitively, cranking the straps tighter often makes leak worse, because it distorts the cushion and creases the seal. Tighten only until the leak stops.
Your mouth opening. If you use a nasal mask or pillows and your mouth falls open in sleep, air escapes straight out — and your machine counts it as leak. Dry mouth in the morning is the giveaway.
Position and pillows. Side sleepers in particular push the mask against the pillow and break the seal. A CPAP pillow, a different mask style, or simply refitting the mask lying on your side can fix it.
AHI: is the therapy actually working?
AHI stands for Apnoea–Hypopnoea Index: the average number of breathing pauses (apnoeas) and periods of shallow breathing (hypopnoeas) per hour. Some apps simply call it "events per hour". It's the same thing.
You'll have met it before your diagnosis. Your sleep study measured your untreated AHI, and it's broadly how severity is graded:
| Untreated AHI (at diagnosis) | Usually described as |
|---|---|
| Under 5 events per hour | Within the normal range |
| 5 to 15 | Mild |
| 15 to 30 | Moderate |
| Over 30 | Severe |
The number on your machine is different
The AHI your CPAP reports each morning is your residual AHI — the events still happening while you're on treatment. For most people, effective therapy brings it down dramatically, and a treated figure under 5 is a commonly used sign that things are well controlled. But your clinic sets your target, not a website, and some people have good reasons for a different one.
Three things to know before you worry about it
It's an estimate. Your machine works out events from changes in airflow. It doesn't know whether you were asleep, and it can't see your oxygen levels. It's a good indicator, not a measurement in the sense your sleep study was.
It varies night to night. Alcohol, a cold, a blocked nose, sleeping on your back, sleeping tablets and simply a bad night can all push it up. A single high reading is rarely meaningful. A sustained change over a couple of weeks is.
Leak can distort it. When leak is high, the machine has a harder job telling a genuine breathing event from air escaping round the mask. If AHI and leak have both risen together, fix the leak first and see whether the AHI follows it down.
When AHI is worth a call
Contact your clinic if your AHI has risen and stayed higher for a couple of weeks with a good seal, if it's high and your daytime sleepiness has come back, or if your report starts showing a lot of events labelled as central or clear airway. Those are a different kind of pause — where the airway is open but the breathing effort briefly stops — and they're something for your clinician to look at, not something to troubleshoot yourself.
What you should never do is change your pressure to bring the number down. Pressure is prescribed, and adjusting it without your clinic is how people make their therapy worse while their data appears to improve.
Mask on/off: the quiet clue.
This counts how many times the mask came off during the night. One or two is normal — most people get up for the bathroom. A larger number tells you something is waking you, and the useful question is what.
Pay particular attention if you find a higher count than you remember. Taking the mask off in your sleep, without waking properly, is common, especially in the first months. It usually means the mask is uncomfortable enough to bother you even when you're not aware of it — too tight, too dry, too warm, or leaking into your eyes.
The fixes follow the cause. Dryness points to humidity. Leak into the eyes points to the cushion or the fit at the bridge of the nose. Discomfort at the cheeks or nose points to over-tightening or the wrong cushion size. And if you're removing it because you feel smothered, that's a genuine and very common reaction with its own set of solutions — our guide on feeling trapped in your mask covers it.
What's actually inside your myAir score.
If you use a ResMed machine with the myAir app, you get a nightly score out of 100. It's built from those same four numbers — but not equally.
On a phone, swipe sideways to see the whole chart.
Look at the proportions and the design becomes obvious: the score is mainly a measure of whether you wore the mask. Usage is worth up to 70 points; everything else together is worth 30. ResMed built it that way deliberately, because consistent use is the thing most strongly tied to therapy actually helping.
That has a practical consequence worth understanding. Your score can look excellent while one of the smaller numbers is quietly getting worse — a long night with a mediocre seal still scores well. So the headline number is a good motivator, but it's worth tapping into the detail now and then rather than only looking at the total.
Pattern, likely cause, first thing to try.
Most problems show up in your data as one of a handful of recognisable patterns. Here's how to read them — and where to start.
| What your data shows | Most likely cause | Try first |
|---|---|---|
| Leak creeping up over weeks | Worn cushion | Replace the cushion; check its age against your replacement schedule |
| Leak still high with a new cushion | Stretched headgear, or a cracked frame | Check strap stretch and flex the frame for hairline cracks |
| Leak spikes, but only some nights | Sleeping position, or the pillow pushing the mask | Refit lying on your side; consider a CPAP pillow |
| High leak plus dry mouth (nasal mask) | Mouth falling open in sleep | A chin strap test, or a full face mask |
| Usage dropping, mask off at 3–4am | Discomfort, dryness, or leak into the eyes | Loosen the straps, adjust humidity, refit at the nose bridge |
| AHI up after changing mask | New mask not sealing, or wrong size | Check the leak figure; refit; ask about sizing |
| AHI up with leak also up | Leak making events harder to read | Fix the leak first, then reassess |
| AHI up with a good seal, sustained | Something clinical has changed | Contact your clinic — don't adjust pressure |
| Great numbers, still exhausted | Something the machine can't see | Tell your clinic; trust how you feel |
The pattern people miss most
Leak that rises slowly over a month and a usage figure that drifts down alongside it. Separately, each change is small enough to ignore. Together they're the classic signature of a cushion reaching the end of its life: the seal weakens, you tighten the straps to compensate, the mask becomes uncomfortable, and you start taking it off earlier. It's very common, entirely fixable, and the data shows it weeks before most people notice.
A silicone cushion is typically a one-to-three-month part. Our replacement schedule covers every part's interval, and the cleaning guide covers how to get the long end of each one.
A five-minute weekly check.
Looking every morning tends to cause worry over nights that don't matter. Once a week, looking at the whole week, is the right rhythm.
Open your app or your machine's history and look at the last seven nights together. Single bad nights happen to everyone — a late night, a cold, a glass of wine. What you're looking for is direction: is anything steadily better or steadily worse than last week?
Any missed nights? Any nights noticeably shorter than usual? If so, think about why while you still remember. If short nights cluster on particular days of the week, that's often a routine issue rather than an equipment one.
Compare this week's leak with a month ago. A slow upward trend is the single most useful early warning in all of your data — it usually means a cushion is wearing out. Note the date you last replaced the cushion and the headgear somewhere you'll find it.
Is it broadly where it usually is? Small fluctuations are normal. A rise that has lasted more than a week or two, especially with a good seal, is worth mentioning to your clinic.
Alert? Sleepy? Better or worse than last month? Write one line. Over time those lines are as valuable as the machine's numbers, and they're exactly what your clinic wants to hear about at review.
Going deeper: SD cards and OSCAR.
The app shows a summary. Many machines also record far more detailed data — and there's a free way to look at it.
OSCAR is free, open-source software, built and maintained by volunteers, that reads the detailed data many CPAP machines record and displays it as graphs: leak through the night, when each event happened, how pressure behaved. For a curious, technically minded user it can be genuinely illuminating — you can see, for example, that your leak spikes at exactly the time you usually roll onto your side.
Three cautions, though. It isn't a medical device, and it isn't validated for clinical decisions. It shows you a great deal more than you need, which can make people anxious about patterns that don't matter. And it should never become a reason to change your own pressure or settings — those are prescribed for a reason, and changing them is a conversation with your clinic.
If you do use it, the most useful thing you can do with it is take a screenshot of a pattern you're concerned about to your next appointment. Clinicians are generally happy to look, and it can make a review far more specific.
When the data says "call your clinic".
Most data problems are equipment problems you can fix at home. These ones aren't — they're for your clinician.
If you drive, your data matters in one more way. At your licence review your clinic forms a view on whether your sleepiness is controlled, partly from this same record. Our guide to sleep apnoea and the DVLA explains how that works.
Six things people get wrong about their CPAP data.
Quick answers on CPAP data.
What is a good AHI on CPAP?
A treated AHI under 5 events per hour is commonly used as a sign of good control, but your clinic sets your personal target. The number on your machine is an estimate that varies from night to night, so look at the trend over a couple of weeks rather than any single reading.
How is the myAir score calculated?
It's out of 100 points: usage hours can earn up to 70, mask seal up to 20, events per hour up to 5, and mask on/off up to 5. Because usage carries most of the weight, the score is mainly a measure of how well you used your machine.
Why is my CPAP leak rate so high?
The most common causes are a worn cushion, stretched headgear, over-tightened straps, your mouth falling open with a nasal mask, and the pillow pushing the mask out of position. If leak has risen gradually over weeks, start by replacing the cushion; if it stays high with a new cushion, check the headgear and frame.
How many hours a night should I use CPAP?
All of your sleep, every night, including naps. The often-quoted four hours is an American insurance compliance threshold rather than a clinical target, and apnoea returns as soon as the mask comes off.
Is the AHI on my CPAP machine accurate?
It's a useful estimate rather than a measurement like a sleep study. The machine works out events from airflow, can't tell whether you were asleep, and can be thrown off by high leak. Treat sustained changes as worth discussing with your clinic rather than relying on any single night.
What does mask on/off mean in myAir?
It counts how many times your mask came off during the night. Once or twice is normal, usually for the bathroom. A higher count, especially if you don't remember it, often means discomfort, dryness or leak is disturbing your sleep.
Is it safe to use OSCAR to look at my CPAP data?
Viewing your data with OSCAR is fine, but it isn't a medical device and isn't validated for clinical decisions. Use it to spot patterns and take screenshots to your clinic, and never use it as a reason to change your own pressure or settings.
More UK-specific CPAP guidance.
This guide is general information, not medical advice. The figures your CPAP machine and app report are estimates intended to support your therapy, not diagnostic measurements. Targets and thresholds mentioned here are general guidance only — your sleep clinic sets the targets that apply to you, and its advice takes precedence over anything on this page.
Never adjust your prescribed pressure or machine settings yourself, whatever your data shows. If your numbers change in a way that concerns you, if daytime sleepiness returns, or if you feel no better despite consistent use, contact your sleep clinic. If you drive and your sleepiness has returned, do not drive until you have spoken to your clinician.
myAir is a ResMed service; the score breakdown described here is taken from ResMed's published guidance at the time of writing and may change. OSCAR is independent free software and is not a medical device.
