Blue lips or fingertips, gasping or choking that does not settle, confusion, chest pain or severe breathlessness need emergency care now, call 911.
Everything below explains how overnight oxygen behaves, what lower numbers can reflect, how the measurement is actually taken, and which bloodwork patterns are worth raising with a clinician.
Blood oxygen is the share of your hemoglobin that is carrying oxygen at a given moment, written as SpO2 and expressed as a percentage. Hemoglobin is the iron-rich protein packed inside red blood cells, and each molecule has four seats for oxygen, so saturation describes how full those seats are, not how much oxygen your blood can carry in total.
A pulse oximeter clips over a fingertip or earlobe, or sits against the wrist or finger in a wearable, and shines red and infrared light through tissue. Oxygen-rich and oxygen-poor hemoglobin absorb those wavelengths differently, and the device reads the pulsing part of the signal to estimate arterial saturation. It is an estimate rather than a direct measurement. The reference method is an arterial blood gas sample drawn by a clinician, which reports both saturation and the pressure of oxygen dissolved in blood.
One detail explains why modest overnight falls get attention. The relationship between oxygen pressure and saturation is a curve, not a straight line. Near the top the curve is flat, so saturation can hold in the mid 90s while oxygen pressure drifts downward. Below roughly 90 % the curve steepens, and each further step down in oxygen pressure costs more saturation than the step before it. For that reason, 88 % and 96 % are not simply eight points apart in physiological terms.
Saturation also says nothing about capacity. Someone with very few red cells can read 98 % and still deliver less oxygen per minute than their tissues want, which is why oxygen readings and blood counts answer different questions and are best read together.
Most healthy adults hold an average overnight saturation in the mid to high 90s, usually within a point or two of their awake value, with short dips that recover on their own. Sleep is not meant to be a flat line, and a single low second on a wearable graph is not a result.
| Overnight SpO2 reading | How it is generally regarded |
|---|---|
| 95 to 100 % | Typical for a healthy adult near sea level, awake or asleep |
| 91 to 94 % | Below typical, often influenced by device fit, altitude or breathing during sleep, worth raising with a clinician |
| 90 % or below | Low, the clinical term is hypoxemia, worth discussing with a clinician promptly if it repeats across nights |
| Brief dips lasting seconds | Common in REM sleep, judged by how deep, how long and how often they occur rather than by one value |
These are general interpretation bands used in clinical practice, not diagnostic cut-offs, and they shift with altitude, age, device accuracy and existing heart or lung conditions. A clinician reads a number alongside symptoms and the shape of the whole night.
That shape is what overnight recordings actually summarise. Four numbers do most of the work. Average saturation across the night describes the baseline. The lowest value, sometimes called the nadir, describes the worst moment. The oxygen desaturation index counts how many times per hour saturation falls by a set amount. Time below 90 %, often shortened to T90, describes how much of the night was spent under that threshold. A night with an average of 96 % and forty dips per hour tells a very different story from a night with a flat 93 % and no dips at all, which is why an average on its own is a weak summary of overnight oxygen.
Small dips are built into normal sleep physiology, not a sign that something has gone wrong. As you fall asleep, the drive to breathe softens. The brainstem becomes less responsive to rising carbon dioxide and falling oxygen, breathing becomes slightly shallower, and saturation settles a touch lower than it sits during the day.
Posture adds to it. Lying flat pushes the abdominal contents against the diaphragm and reduces the volume of air left in the lungs at the end of a quiet breath. That reserve is what buffers saturation between breaths, so a smaller reserve means a faster fall when a breath is shallow or delayed.
REM sleep is the most demanding stretch of the night. During REM, most skeletal muscles lose their tone, including the muscles of the chest wall and the muscles that hold the upper airway open. Breathing becomes irregular and leans heavily on the diaphragm. Dips are deeper and a little longer in REM than in any other stage, which is why many people see their worst numbers in the second half of the night when REM periods lengthen.
The practical distinction is between dips that are shallow, brief and scattered, and dips that are repetitive, deep and clustered. The first pattern is ordinary. The second suggests that breathing is being interrupted rather than simply relaxed, and that is the pattern worth taking to a clinician.
Repeated, cyclical falls usually mean airflow is being interrupted. In obstructive sleep apnea the upper airway narrows or closes while the effort to breathe continues, airflow stops or drops, oxygen falls, and a brief arousal from sleep reopens the airway, after which the cycle restarts 1. Central events differ in that the effort to breathe pauses as well. A third pattern, hypoventilation, produces a sustained shallow fall rather than a saw-tooth, and can be seen in people with significant obesity, neuromuscular weakness or chronic lung conditions.
The repetitive pattern is what makes the obstructive form distinctive. Research suggests that intermittent hypoxia, the repeated fall and rapid rise in oxygen across a night, together with sympathetic nervous system activation and inflammatory signalling, sits at the centre of how obstructive sleep apnea affects the rest of the body 2. Each cycle is a small stress, and a night can contain hundreds of them.
That matters for how the numbers are read. A single long desaturation at 03:00 may reflect a position change or a device slipping. Forty shallow desaturations an hour, each followed by a quick recovery, is a rhythm rather than an accident. Clinicians describe obstructive sleep apnea by counting these breathing events per hour of sleep and pairing the count with symptoms such as loud snoring, witnessed pauses in breathing and daytime sleepiness 1.
None of this can be settled from an overnight graph alone. The graph raises the question. A clinician decides what the question means for you and whether any next step, including formal sleep testing, is appropriate.
Hematocrit is the percentage of your blood volume made up of red blood cells, measured as part of a routine complete blood count. Labs regard roughly 36 to 52 % in men and 33 to 48 % in women as typical, and FitScript's optimal range is 40 to 50 % in men and 36 to 44 % in women. Optimal ranges are FitScript's optimization targets, not diagnostic cut-offs, and standard reference ranges vary by lab, age and sex. The full picture of what the marker is, how it is tested and how a result is interpreted lives on the Hematocrit guide. It earns a place here for one reason. Red cell production answers to oxygen supply, so a night-time breathing problem can leave a fingerprint in a morning blood draw.
Sustained low oxygen is the body's signal to build more oxygen carriers. Oxygen sensors in the kidney detect a shortfall and increase the release of erythropoietin, a hormone that tells bone marrow to produce more red blood cells. Over weeks, more circulating red cells raise both hemoglobin and hematocrit. This is the same adaptation that happens at high altitude, and it is a compensation rather than a disease in itself.
When the trigger is night-time breathing, the picture is less tidy. The dips are intermittent rather than continuous, and the rise in red cell mass, when it appears, is usually modest rather than dramatic. Research suggests that intermittent hypoxia drives a broad set of downstream effects in obstructive sleep apnea, with sympathetic activation and oxidative stress alongside the oxygen swings themselves 2. A hematocrit sitting at the upper end of the standard range in a heavy snorer who wakes unrefreshed is not a diagnosis. It is a pattern worth putting in front of a clinician alongside the sleep history.
A raised hematocrit has several plausible explanations that have nothing to do with sleep. Dehydration concentrates the blood and can lift the reading on the day of the draw. Living at altitude raises it legitimately. Smoking raises it. Some medical conditions raise it, and a licensed provider is the one who sorts between them. The useful move is not to interpret the number alone but to look at it next to hemoglobin, red cell indices, symptoms and, where relevant, an overnight oxygen recording.
The reverse direction is worth knowing too. A low hematocrit does not protect you from night-time desaturation, and it can make the same drop in saturation harder on the tissues, because there are fewer carriers available in the first place.
Some people start the night with less margin than others. Obstructive sleep apnea is associated with obesity, increasing age, male sex and craniofacial features that narrow the upper airway, and the condition is common enough that many cases go unrecognised for years 3. Nasal obstruction, enlarged tonsils and a crowded oropharynx all reduce the space available when muscle tone falls.
Other groups see lower overnight numbers for different reasons. People with chronic lung conditions often begin the night closer to the steep part of the saturation curve, so a normal REM dip takes them further down than it would take someone with healthy lungs. People with heart failure can show irregular breathing patterns during sleep. Those living at altitude run lower saturations throughout the day and night, which is expected rather than abnormal for the setting.
Women are not exempt, although the presentation often differs. Snoring may be lighter, the main complaint may be fatigue, insomnia or morning headache rather than classic witnessed pauses, and risk changes across the menopausal transition. That mismatch with the stereotype is one reason overnight oxygen data can be useful, because it describes what the night actually looked like rather than what a bed partner noticed.
Sleeping position matters as well. Many people desaturate more while on their back than on their side, which is why a single-night recording can look worse or better than usual depending on how the night happened to go.
Plenty. Pulse oximetry depends on a clean pulsatile signal through tissue, and anything that weakens that signal degrades the reading. Cold hands and poor peripheral circulation are the most common culprits, because a constricted fingertip gives the device very little to work with. Movement during sleep creates artefact, and a wrist-worn sensor that shifts loose during the night can report falls that never happened in the bloodstream.
Other interference is physical. Nail polish and artificial nails can affect light transmission through a fingertip sensor. Strong ambient light reaching the sensor can disturb it. Skin pigmentation, tattoos and certain dyes can affect accuracy, and the same device can perform differently on different people. Carbon monoxide exposure is a specific trap, because standard oximeters can read high while oxygen delivery is impaired.
Altitude is not an artefact but it is a context. Saturation legitimately falls as you go higher, so a reading taken on a mountain trip is not comparable to one taken at home.
The practical rule is to distrust an isolated number and trust a pattern. One dip to 88 % on one night, with a loose strap and cold hands, is weak information. The same dip appearing every night, clustered in the small hours, with morning headache and daytime sleepiness alongside it, is a reason to book a conversation with a clinician rather than to keep collecting graphs.
Through pulse oximetry, recorded continuously or at intervals across the night, with the amount of surrounding detail depending on the method used. The options sit on a ladder from casual to clinical.
| Method | What it records | Where it generally fits |
|---|---|---|
| Consumer wearable (watch or ring) | Periodic SpO2 estimates, heart rate, movement | Spotting patterns across many nights, never a diagnosis |
| Overnight pulse oximetry | Continuous saturation and pulse, with summary metrics for the night | Often arranged by a clinician to quantify how often and how deeply oxygen falls |
| Home sleep apnea test | Airflow, breathing effort, oxygen, pulse | Clinician-arranged testing in suitable cases |
| In-lab sleep study (polysomnography) | Brain activity, sleep stages, breathing, oxygen, limb movement | The fullest assessment, ordered by a clinician when detail or complexity demands it |
The in-lab study remains the reference, because it records sleep itself rather than inferring it, and it counts breathing events against actual sleep time 1. Simpler recordings trade detail for accessibility and a night in your own bed.
One point about interpretation is easy to miss. A home oximeter shows oxygen but not the reason it fell. Airflow and effort channels are what separate an obstructive event from a central one, and brain activity is what distinguishes true sleep from lying still with your eyes closed. That is why a low home reading leads to a clinical conversation rather than to a conclusion, and why a licensed provider decides which level of testing, if any, is appropriate for a given person.
Accurate enough to be informative, not accurate enough to settle anything on their own. Oximetry-based devices have been examined specifically as a route to identifying obstructive sleep apnea, and a systematic review and meta-analysis of these devices suggests they can detect the condition with useful diagnostic performance compared with reference testing 4. Performance is generally strongest for more severe disease and weaker for mild disease, where the oxygen dips are shallower and easier to miss.
Algorithmic analysis has pushed this further. Research suggests that machine learning models applied to overnight pulse oximetry signals can identify obstructive sleep apnea with promising accuracy, as summarised in a systematic review and Bayesian meta-analysis of these approaches 5. That is a statement about what the recordings contain, not an invitation to self-diagnose from an app summary.
There is also a practical gap between a medical-grade overnight oximeter and a consumer wearable. A dedicated recorder samples continuously from a fingertip probe. Many wearables sample intermittently from the wrist, skip stretches of the night, and apply their own smoothing before showing you a chart. Values below 90 % on wrist devices should be treated as a prompt to ask, not as a measurement to act on.
The sensible way to use a wearable is as a reason to start a conversation. If several nights show repeated dips, bring the data to a clinician and let them decide whether a validated recording or a sleep study is the right next step.
Because the rest of the body is awake to what happens at night. Each obstructive event ends with an arousal and a surge of sympathetic activity, and that cycle repeats for hours. Obstructive sleep apnea is associated with hypertension, including forms of high blood pressure that respond poorly to standard management, and the relationship runs in both directions 6.
The stress shows up in other settings too. Research suggests that obstructive sleep apnea is associated with a higher rate of cardiac complications after surgery, which is one reason anaesthetists screen for it before an operation 7. Daytime consequences are more familiar, with excessive sleepiness, impaired concentration and unrefreshing sleep among the usual complaints 1.
Underneath these associations sits the same repeated oxygen swing described earlier, along with the inflammatory and oxidative changes that accompany it 2. The hematocrit story belongs in that frame as well. A rising red cell percentage is one of the body's slower answers to a nightly oxygen shortfall, which is what makes it an interesting marker to track rather than a number to panic about.
None of this means a low reading on a wearable predicts your future. It means night-time breathing is a legitimate health variable, it is measurable, and it is worth including in the same conversation as blood pressure and bloodwork.
The blood count comes first, because it reflects oxygen carrying capacity and its adaptations. Hematocrit and hemoglobin are the headline numbers, with red cell count, mean corpuscular volume and red cell distribution width adding texture about cell size and uniformity. Iron studies, including ferritin, give context when hemoglobin is low, since a shortage of raw material limits production regardless of oxygen signalling.
After that, the markers that describe cardiometabolic load are the most relevant companions. Fasting glucose and HbA1c, a lipid panel, kidney function and an inflammatory marker such as hs-CRP build a picture of the terrain that disrupted sleep acts upon. Thyroid function is often checked when unexplained fatigue is part of the story. Blood pressure, while not a lab test, belongs on the same page.
This is where repeat testing earns its keep. A hematocrit of 49 % means little in isolation and quite a lot if it was 44 % a year ago and the person has started snoring. FitScript's membership is built around that kind of tracking, with 150+ biomarkers tested per year and interpretation by Atlas, FitScript's AI health advisor, so changes are read as trends rather than as single snapshots.
Testing has limits worth naming. Bloodwork cannot tell you how many times you stopped breathing last night, and an overnight oxygen trace cannot tell you whether your iron stores are adequate. They are complementary, and a clinician is the one who decides which combination answers your particular question.
Several everyday variables move the numbers, which is why one night is a poor sample. Body weight is among the strongest, since obesity is associated with obstructive sleep apnea and with narrowing of the upper airway during sleep 3. Changes in weight in either direction can shift overnight recordings over months.
Shorter-term factors move things faster. Nasal congestion from a cold or allergies increases the effort needed to move air. Alcohol in the evening relaxes upper airway muscles and is included among the recognised contributors to obstructive sleep apnea. Sleeping on your back rather than your side deepens dips for many people. Travel to altitude lowers baseline saturation for as long as you stay there. A chest infection lowers it while the infection lasts.
Life stage matters too. Pregnancy changes lung volumes and airway tissue. The menopausal transition changes both sleep architecture and airway risk. Normal ageing softens the ventilatory response and loosens airway tissue.
This guide does not prescribe what to do about any of these, because that decision belongs with a licensed provider who can see your history. What the list is good for is interpretation. If your overnight numbers worsened during a fortnight of heavy congestion and late drinks on a work trip, the recording may be describing that fortnight rather than describing you. If the trend holds across months of ordinary nights, it is describing something more durable.
Whenever the pattern repeats, or whenever symptoms point the same way the numbers do. Loud snoring, witnessed pauses in breathing, gasping awakenings, morning headaches, unrefreshing sleep and daytime sleepiness are the classic cluster, and they are the features clinicians use alongside testing when assessing obstructive sleep apnea 1. Add a hematocrit drifting toward or above the top of the standard range, and the conversation gets more concrete.
Readings of 90 % or below that recur across several nights are suboptimal and worth discussing with a clinician promptly, even when you feel fine during the day. The same applies to new breathlessness lying flat, ankle swelling, chest discomfort at night or a sudden change in exercise tolerance. Never ignore a result that sits outside the standard range because a device might be imperfect, and never assume that feeling well rules out a problem that only happens while you are unconscious.
Bring the raw material to the appointment. Several nights of recordings rather than one, a note of what the nights involved, recent bloodwork, your blood pressure readings, and a description of how you feel on waking. That package lets a clinician decide whether formal sleep testing is warranted and what, if anything, follows it. Any management option, including anything requiring a prescription, is evaluated by a licensed provider who has reviewed you in person.
This article is educational and is not medical advice. Discuss your own results and any treatment decision with a clinician or licensed provider.
For related reading, see whether can sleep apnea cause weight gain describes the loop you are caught in, and if your worst nights cluster at a particular point in the month, trouble sleeping before period covers that pattern in more detail.
Most healthy adults hold an overnight average in the mid to high 90s, usually within a point or two of their awake reading. Brief dips lasting seconds are ordinary, especially during REM sleep. Readings that sit at or below 90 % repeatedly across nights are below typical and worth raising with a clinician.
Yes. During REM, most muscles lose tone, breathing becomes irregular and lung volumes are smaller when lying flat, so saturation dips more than at other stages. Normal dips are shallow, short and scattered. Dips that are deep, repetitive and clustered are a different pattern and deserve a clinical opinion.
On its own it means a device recorded a low value, not that you have a condition. A single reading can come from a loose sensor, cold hands, movement or a position change. If 87 % shows up repeatedly on several nights, that is suboptimal and worth discussing with a clinician promptly so they can decide whether formal sleep testing is appropriate.
No. Wearables estimate saturation at intervals from the wrist or finger and apply their own smoothing, so they are useful for spotting patterns across many nights rather than for making a diagnosis. Only a clinician can diagnose a sleep breathing disorder, usually after a home sleep test or an in-lab sleep study.
A sustained shortfall of oxygen prompts the kidneys to release more erythropoietin, which tells bone marrow to make additional red blood cells. Over weeks that can lift hemoglobin and hematocrit. A higher hematocrit has several other explanations too, including dehydration, altitude and smoking, so a clinician interprets it in context.
Not reliably. Some people have frequent breathing interruptions with only modest oxygen dips, particularly in milder disease, and intermittent wrist measurements can miss short events entirely. Symptoms such as loud snoring, witnessed pauses, morning headaches and daytime sleepiness matter as much as the number, so mention them even when readings look fine.
Cold fingers, poor circulation, movement, a loose strap, nail polish, bright ambient light and sensor placement all degrade the signal. Altitude lowers saturation genuinely rather than artificially. Warm your hands, secure the sensor and compare several nights before drawing any conclusion from one chart.
Book when readings at or below 90 % repeat across nights, or when snoring, gasping awakenings, morning headaches or persistent daytime sleepiness accompany the numbers. Bring several nights of recordings, recent bloodwork and your blood pressure readings. Blue lips, confusion, chest pain or severe breathlessness need emergency care straight away.