Agonal breaths which sound like isolated or infrequent gasps or snoring occur in roughly 30 to 40 percent of out-of-hospital cardiac arrests, with most large series reporting about 33 to 39 percent; in witnessed arrests the figure climbs toward 40 percent, while bystander-reported "normal breathing" coexists with agonal gasps in a substantial minority of cases. The range reflects how the sign is defined and who is counting: emergency medical services (EMS) crews using formal criteria report lower rates than lay witnesses describing what they saw.
The number matters because agonal breathing is the single most common reason a bystander decides not to start cardiopulmonary resuscitation (CPR). A gasping patient looks alive. That misread costs minutes, and minutes cost survival.
What an agonal breath actually is
An agonal breath is not a breath. It is a brainstem reflex that fires when higher cortical control of breathing has already failed. The medulla contains a primitive pattern generator that can trigger a gasping inspiratory effort when it detects catastrophic hypoxia or hypoperfusion, independent of any conscious drive to breathe. This is why the movement can appear in a patient who is pulseless and clinically dead.
The reflex produces a characteristic motor pattern. The mouth opens, the jaw thrusts, the neck extends, and the chest may rise sharply, sometimes with a snoring, gurgling, or rattling sound from soft tissue vibration in the upper airway. The rate is slow and irregular, typically fewer than 8 per minute, often 2 to 4, and the efforts come in clusters separated by long silences. There is no effective tidal volume. Alveolar ventilation during agonal respiration is functionally negligible, so oxygen saturation falls despite the chest movement.
The reflex is self-limiting. In animal models and observational human data, agonal breathing persists for a median of about 2 to 4 minutes after circulatory arrest and rarely beyond 10 minutes without CPR. That window overlaps almost exactly with the interval in which defibrillation and chest compressions are most effective. The gasp is therefore a timing signal, not a sign of life.
One level deeper: the gasps are generated by the pre-Bötzinger complex and neighboring medullary neurons that retain intrinsic pacemaker activity after cortical and pontine input is lost. Because these neurons depend on residual blood flow, the reflex fades as brainstem ischemia deepens. A patient who stops gasping has not "improved"; the brainstem has gone quiet. This is the counterintuitive point that reverses the lay intuition: continued gasping indicates a brainstem that is still perfused enough to fire, which is a better prognostic sign than silence.
Quantifying how often it happens
The 30 to 40 percent figure comes from converging sources rather than a single registry. EMS- and dispatcher-based studies that audited recordings or used structured witness interviews have reported agonal respiration in roughly one-third of out-of-hospital cardiac arrests, with ranges from about 25 percent to just over 40 percent depending on the case mix. In-hospital data are sparser and generally lower, because arrest is more often witnessed by clinicians who recognize pulselessness immediately.
| Data source type | Reported frequency | What drives the difference |
|---|---|---|
| Dispatcher-audited out-of-hospital series | ~30–40% | Structured interview, trained recognition |
| Lay bystander self-report | ~25–50% | Recall bias, loose definition of "gasping" |
| EMS clinician on-scene documentation | ~20–30% | Under-recording, focus on pulselessness |
| In-hospital witnessed arrest | ~10–20% | Immediate clinician recognition |
| Arrests with bystander CPR already started | ~35–45% | Longer pre-EMS interval, reflex more likely captured |
The pattern across rows is consistent: the longer the interval between collapse and professional assessment, the more likely a witness will observe at least one gasp. That is a function of the reflex's natural time course, not of geography or technique.
Plain-language version: about one in three cardiac arrest patients will gasp, snore, or make noisy breathing movements at some point before EMS arrives.
When the numbers mislead: edge cases
Several conditions inflate or deflate the observed percentage, and each one changes what a witness should do.
Opioid-associated arrest behaves differently. In respiratory arrest driven by opioid toxicity, agonal gasps can appear earlier and last longer because the heart may still be beating when breathing fails. This is the clinical basis for the recommendation that lay rescuers give naloxone and rescue breaths before compressions in suspected opioid overdose with a pulse. The reflex here signals respiratory failure, not circulatory arrest, and the treatment sequence differs.
Hypothermic arrest stretches the timeline. Cold-water immersion can produce agonal breathing that persists far longer than the 2 to 4 minute norm, and the same neuroprotective effect that prolongs the gasp also improves the odds of successful resuscitation after prolonged CPR. A gasp in a cold-water drowning victim is a reason to continue, not to stop.
Pediatric arrest is dominated by respiratory causes. Agonal breathing in children is more frequent than in adults and more often precedes full circulatory collapse, which is why pediatric basic life support teaches rescue breaths as a primary step.
Terminal gasps near expected death are a separate phenomenon. In hospice and palliative settings, agonal breathing in the final minutes is anticipated and is managed by positioning and medication, not by CPR. The reflex is the same; the appropriate response is not.
The professional caveat: the 30 to 40 percent figure is a population estimate, not a bedside rule. A witness who sees one gasp should not calculate odds. They should check for responsiveness and normal breathing for no more than 10 seconds, and if absent, start CPR.
The misconceptions that cost lives
The most damaging misconception is that gasping equals breathing. Lay rescuers in dispatcher-assisted CPR studies frequently describe agonal respiration as "breathing" or "snoring" and delay compressions by several minutes. Dispatcher protocols now coach callers to treat any irregular, slow, or noisy breathing in an unresponsive adult as cardiac arrest.
A second misconception is that the presence of a gasp means a pulse exists. It does not. The reflex is generated centrally and does not require cardiac output. Pulse checks by lay rescuers are unreliable even when attempted, which is why current basic life support guidance removes them from the lay algorithm entirely for adult arrest.
A third is that gasping is a sign of recovery during CPR. It can be, but only in a specific pattern: gasps that become more frequent and regular while compressions are underway may indicate return of spontaneous circulation. Isolated gasps in an unresponsive patient with no other signs do not.
Red-flag warnings:
Do not delay compressions to "let the patient breathe" when gasping is the only respiratory effort. Do not assume a snoring sound means an obstructed airway in an unresponsive patient; treat it as arrest until proven otherwise. Do not stop CPR because the patient gasped once; reassess only per protocol or when exhausted. Do not give anything by mouth to a gasping patient; the airway is unprotected. Do not mistake post-ictal stertor after a seizure for agonal breathing if the patient is otherwise recovering; context and a pulse check by a professional distinguish them.
What to do with a gasp: thresholds and actions
The practical response is algorithmic and does not require knowing the percentage.
Check responsiveness with a shout and a shoulder tap for no more than 10 seconds. Look for normal breathing, not any breathing. Normal means regular, quiet, and adequate; agonal gasps, snoring, and isolated sighs do not count. If unresponsive and not breathing normally, call emergency services or direct someone to call, and start chest compressions at 100 to 120 per minute with a depth of at least 2 inches in adults. Send for an automated external defibrillator (AED) and apply it as soon as it arrives; do not interrupt compressions to fetch it alone. Continue until EMS takes over, an AED advises a shock and you deliver it, or the patient shows clear signs of life such as purposeful movement and normal breathing. In suspected opioid overdose with a pulse but inadequate breathing, give naloxone and rescue breaths per local protocol before compressions.
The 10-second threshold matters. Longer breathing checks are associated with delayed compressions and worse outcomes, and they do not improve recognition of agonal breathing because the pattern is intermittent by nature.
Time course and why timing changes the response
| Interval after collapse | Typical agonal pattern | Practical implication |
|---|---|---|
| 0–30 seconds | First gasps, often 2–4 efforts | Witness may still see the patient as alive |
| 30 seconds–2 minutes | Irregular clusters, snoring sounds | Highest-risk window for bystander delay |
| 2–4 minutes | Median duration of persistence | Overlaps best window for defibrillation |
| 4–10 minutes | Reflex fades in most patients | Silence does not mean recovery |
| Beyond 10 minutes | Rare without CPR | Brainstem ischemia deepens |
| With effective CPR | Gasps may continue or return | Not a reason to pause compressions |
| After return of circulation | Gasping may precede regular breathing | Reassess per protocol, do not assume stability |
The table's central lesson is that the reflex is a clock, not a status. Its presence tells you the arrest is recent; its absence tells you little about viability.
Limits of the evidence
The 30 to 40 percent estimate is derived largely from observational series with heterogeneous definitions. Some studies count any gasp, others only gasps recorded by EMS, and dispatcher audits depend on caller description quality. There is no national registry that codes agonal breathing as a standalone variable with uniform criteria, so the true population frequency may sit anywhere from about one-quarter to nearly one-half of arrests depending on ascertainment.
Recall bias runs in both directions. Bystanders under stress may fail to report a brief gasp, inflating false negatives, or may label any noisy breathing as agonal, inflating false positives. Studies using audio recordings reduce but do not eliminate this problem.
What can be concluded with confidence is directional and stable: agonal breathing is common, it is frequently mistaken for normal breathing, and that mistake delays CPR. The precise percentage is less important than the recognition rule it supports. Any unresponsive patient who is not breathing normally needs compressions, regardless of whether the breathing looks like a gasp, a snore, or a sigh.
This article is general educational information about a physiologic sign, not a diagnosis, treatment plan, or instruction to substitute for training. Do not use it to make decisions about a specific patient; call emergency services and follow dispatcher guidance, and seek hands-on certification from a qualified local provider. In an emergency, contact local emergency services immediately.