How to Read Your CPAP Data: A Plain-English Guide to AHI, Leak Rate & Pressure

Your CPAP machine records a surprising amount of data every night — far more than the cheerful "great job!" badge on your phone app suggests. The trouble is that the most useful numbers are either buried in a summary screen or locked away on an SD card. This guide walks you through what your machine actually tracks, the four numbers worth knowing, and how to read them in plain English without becoming an OSCAR power user.

What your CPAP actually records

There are two very different layers of data inside your machine, and knowing which one you're looking at changes everything.

The summary layer is what your manufacturer's app shows you. ResMed's myAir, the now-discontinued Philips DreamMapper (it shut down in January 2026, with users pushed to migrate elsewhere), and similar apps display a handful of nightly totals: how long you used the machine, an estimated AHI, a leak figure, and a friendly score. This data syncs over the cellular modem or Wi-Fi built into modern machines.

The detailed layer lives on the SD card inside the machine. This is the breath-by-breath record — every flagged event, the flow-rate waveform (a tracing of your actual breathing), pressure changes second by second, and leak measured across the whole night. This is the same data your sleep clinician's software reads, and it's the layer that actually lets you understand what happened.

The gap between these two layers is the central frustration of CPAP self-monitoring. The app gives you a verdict; the SD card gives you the evidence.

The four headline numbers — AHI, leak rate, 95th-percentile pressure, usage hours

Most of what you need to know night-to-night lives in four numbers. Here's what each one means and the rough benchmark to keep in mind.

AHI (Apnea-Hypopnea Index) counts breathing interruptions per hour: apneas (airflow stops) and hypopneas (airflow drops). The widely used treatment goal for adults on CPAP is a residual AHI below 5 events per hour. The AASM defines an "optimal" PAP titration as reducing AHI to fewer than 5, and below 5 is also the normal, non-apneic range. Some clinicians aim lower for fuller symptom control when it's comfortably achievable, but there's no formal guideline establishing "below 1" or "below 2" as a recognized standard — and targets are individualized. Someone with severe OSA who lands at a residual AHI of 6 may still be doing well. For a fuller breakdown, see what is a good AHI on CPAP.

One important caveat: the AHI your machine reports is an estimate. It's calculated from airflow and pressure signals alone. A lab sleep study (polysomnography) scores events using EEG brain-wave data and blood-oxygen measurement, so device-reported AHI and lab-scored AHI will not match exactly. Your machine can't see arousals it didn't cause a pressure response to, and it has no oxygen sensor.

Leak rate is the most underrated number on the report — and the one that can quietly invalidate everything else. When air escapes around your mask or through your mouth, the machine's airflow sensor gets confused. Large leak can cause your machine to under-report AHI, which means a reassuringly low AHI on a high-leak night can be misleading. Always glance at leak before you trust the AHI. ResMed flags trouble at roughly 24 L/min of excess leak at the 95th percentile. (A wrinkle: ResMed reports excess leak above an expected baseline, while Philips reports total leak — so the raw numbers aren't directly comparable between brands.) See acceptable CPAP leak rate for thresholds and fixes.

95th-percentile pressure is the pressure level your machine stayed at or below for 95% of the night. On an auto-adjusting machine (APAP), this tells you what pressure your airway actually demanded most of the night, ignoring brief spikes. It's far more useful than the maximum pressure, which might reflect a single restless moment. The 95th-percentile vs median pressure guide explains how to read this against your prescribed range.

Usage hours is simply how long the machine ran while you were breathing on it. Insurance and many clinicians use a 4-hours-per-night threshold to define "compliance," but that's a billing floor, not a health target — more sleep on therapy is better. Our usage hours guide covers what "compliant" really means.

Why one night is noise and trends over weeks matter

Here is the single most important habit to build: don't react to a single night. Your AHI naturally bounces around from night to night based on sleep position, alcohol, congestion, how much REM sleep you got, and plain randomness. A one-off AHI of 8 after a string of 2s is almost never a problem — it's noise.

What actually tells you something is the trend over two to four weeks. A slow creep upward, a leak number that climbs every night, or a pressure that's been pinned at your maximum for a week — those patterns are signal. A single bad night is not. If you find yourself alarmed by nightly swings, the night-to-night variation guide puts it in perspective.

Beyond the headline numbers — event types, flow limitation, waveforms

The four headline numbers tell you whether something happened. The detailed SD-card data tells you what and why — and this is where reading your own data gets genuinely powerful.

Event types. Your AHI lumps together different kinds of events that mean very different things:

• Obstructive apneas — your airway physically collapses. This is classic OSA, and it's exactly what CPAP is built to fix by splinting the airway open.
• Central / clear-airway (CA) apneas — your brain briefly doesn't send the signal to breathe. The airway is open, but no effort is made.
• Hypopneas — partial reductions in airflow rather than full stops.
• RERAs — respiratory effort-related arousals, subtle events that disrupt sleep without meeting the apnea/hypopnea threshold.

The obstructive-versus-central distinction matters a lot. CPAP reliably treats obstructive events but does not directly correct the unstable breathing drive behind central apneas. If you see central/clear-airway events rising on CPAP, that can indicate treatment-emergent central sleep apnea (TECSA), sometimes called complex sleep apnea. The good news: TECSA often resolves on its own within weeks to a few months of continued CPAP (reported spontaneous resolution is roughly 60–80%). Management may be watchful waiting on CPAP under your clinician's guidance, or — if it persists — a switch to BiPAP/ASV or other measures. What it is never a reason to do is raise your own pressure to "chase" central events; higher pressure doesn't fix them and can sometimes provoke more. Persistent central events, especially with returning sleepiness, deep oxygen drops, or new medications (like opioids), deserve a clinician's review. The event types decoded guide and our deep dive on central apneas on CPAP (TECSA) cover this in full.

Flow limitation is the metric that hides behind a "good" AHI. It measures subtle flattening of your breaths — your airway narrowing without fully closing — and it doesn't count toward AHI at all. You can have a textbook AHI of 2 and still feel exhausted because of persistent flow limitation or UARS. See flow limitation: the hidden metric if you're still tired despite great numbers.

The flow-rate waveform is the raw breathing tracing — the actual shape of each inhale and exhale. Learning to spot a "flat-top" breath (a sign of flow limitation) or the crescendo-decrescendo pattern of periodic breathing turns you from a number-reader into someone who can see what their airway is doing. Our flow-rate waveform guide is the gentle introduction.

Why your machine's app shows so little

If the SD card holds all this richness, why does the app show four numbers and a smiley face? Because the apps were designed for encouragement and compliance reporting, not analysis.

ResMed's myAir is the clearest example. It works only with ResMed machines, and its headline is a 0–100 score weighted heavily toward usage hours. The exact formula has never been published, and because usage dominates, you can score a perfect 100 with an AHI of 4.9 and meaningful leak. myAir shows no event-type breakdown, no flow-limitation data, and no waveform. It's a motivational tool, not a diagnostic one — see how to read your myAir score for what it hides.

Philips users are in a tougher spot: DreamMapper was shut down in January 2026, and the newer DreamStation 2 writes its SD-card data in an encrypted format that OSCAR and most third-party tools cannot read at all. (SomniCharts is an exception — it does decode DreamStation 2 data. More on that in the next section, and in our DreamStation 2 data guide.)

The pattern across all manufacturer apps is the same: they summarize, they don't explain. The myAir & DreamMapper limitations guide lays out exactly where each falls short.

How to get the full picture without learning OSCAR

The traditional way to read your detailed data is OSCAR — a free, well-respected desktop program. It's genuinely good, but it comes with real friction: you have to physically pull the SD card, buy a card reader, install desktop software (it won't run on a Chromebook), and then learn to interpret raw charts that have no plain-language explanations and no auto-scoring. For many people that's a weekend project they never start. Our OSCAR software guide covers it honestly.

This is exactly the gap SomniCharts was built to close. It auto-imports the same detailed data OSCAR reads — but in the cloud, with no install and no SD-card reader required — and explains it in plain language automatically. It works across ResMed, Philips Respironics (including the encrypted DreamStation 2), and Löwenstein prisma machines, and its built-in assistant, SomniDoc, walks you through what your AHI, leak, pressure, and event patterns actually mean for your therapy. You upload your data, and you get an explanation instead of a wall of charts.

Whichever tool you choose, the goal is the same: turn your data into an informed conversation with your sleep clinician. Reading your own numbers helps you ask better questions and notice trends early. It does not replace your provider's judgment — and it's never a license to adjust your prescribed pressure on your own.

For the bigger picture, start with the pillar guide, Reading Your CPAP Data, and branch out from there into the specific number that's on your mind tonight.

Frequently asked questions

What's a normal AHI on CPAP? A residual AHI below 5 events per hour is the widely used benchmark for effective therapy, matching the AASM's definition of "optimal" titration. Some clinicians aim lower when it's comfortably achievable, but goals are individualized — discuss your specific target with your provider.

Why is my AHI low but I still feel tired? A low AHI doesn't capture everything. Persistent flow limitation, RERAs, fragmented sleep, or large leak invalidating the AHI can all leave you exhausted with "good" numbers. The detailed waveform and flow-limitation data tell the rest of the story.

Can I trust the AHI my machine reports? Treat it as a reliable estimate, not a lab result. Your machine calculates AHI from airflow and pressure alone — it has no EEG or oxygen sensor, so it differs from a lab-scored AHI. And remember that high leak can cause it to under-report, so always check leak first.

Should one bad night worry me? No. Single-night numbers are noise. AHI naturally varies with sleep position, alcohol, and congestion. Watch the trend over two to four weeks — that's what reveals the real pattern.

Do I need to buy an SD-card reader and learn OSCAR? Not necessarily. OSCAR is a capable free desktop tool, but cloud options like SomniCharts auto-import and explain the same data without an install or card reader, and support ResMed, Philips (including DreamStation 2), and Löwenstein machines.