Croup: the barking cough, how doctors tell it apart, and when it is an emergency

Short answer

Croup is swelling of the voice box and windpipe, almost always from a virus, in children roughly 6 months to 3 years old. It is diagnosed by ear and eye — a barking cough, a hoarse voice, and a high-pitched noise breathing in — not by a test (2, 4).

If you are reading this at 2 a.m. with a child who sounds like a seal, you are not imagining it and it is not your fault. The sound is genuinely frightening, it is much worse at night than it was at the pediatrician's office that afternoon, and most parents meet it with no warning at all (2, 12). Croup is common: around 352,000 cases a year in the US, about 1.4% of all children's emergency department visits (4). Most of it is mild and settles in days (3, 12). The job of this article is to help you tell the common version from the few conditions that look like it and are not.

Note: this article explains what guidelines and research say about recognizing croup and is not medical advice; your child's doctor decides what this is and what to do. Call 911 if your child has stridor — the high-pitched noise breathing in — while resting or asleep; is sucking the skin in above, below or between the ribs to breathe; looks blue, grey, pale or blotchy on the lips, tongue or skin; is drooling and cannot swallow; or is limp, unusually sleepy, or not recognizing you. In the UK call 999 (2, 3).

What croup actually is, and why it sounds like that

Croup is "a condition that causes a swelling of the voice box (larynx) and windpipe (trachea)," as the American Academy of Pediatrics puts it (2). The swelling sits in the narrowest part of a small child's airway, just below the vocal cords. Two consequences follow, and they explain everything a parent hears.

First, the cough. Air forced past a swollen, stiffened voice box makes a hard, hollow, metallic sound — the barking seal or the sea lion. It is not a chesty cough and not a wet one. It is unmistakable once you have heard it, which is why clinicians often say they can diagnose croup from the waiting room.

Second, stridor. When the narrowed section of airway has to pull air through on an in-breath, it produces what the AAP describes as "a coarse musical sound each time they breathe in" (2). The NHS calls it "a high pitched, rasping sound when breathing in" (3). Stridor is the single most informative sign in croup, and the thing clinicians grade most carefully — because stridor only when a child is upset means something very different from stridor in a sleeping child.

A small child's airway is the reason croup is a childhood illness. The same millimetre of swelling that a teenager would barely notice takes a meaningful share of a toddler's airway diameter. Croup usually affects children aged about 6 months to 3 years, with an average age around 2.5 years, and it is somewhat more common in boys (4, 5, 6). About 3% of children between 6 months and 3 years get it, and croup accounts for roughly 7% of hospitalizations for fever or acute respiratory illness in children under 5 (5).

Which virus causes it

Most croup is viral, and the usual culprits are human parainfluenza viruses, chiefly types 1 and 3 (6). The CDC states plainly that "HPIV-1 infections often cause croup in children," and that HPIV-2 can do the same (1). Influenza A and B, respiratory syncytial virus, rhinovirus, coronaviruses, human metapneumovirus and adenovirus all turn up too (6). SARS-CoV-2 can cause croup as well, and the Omicron variant has been linked to a more severe course (4).

Because it is a virus, croup is contagious in the ordinary way a cold is contagious — and it spreads before the barking starts. CDC puts the incubation period for parainfluenza at generally 2 to 6 days from infection to symptoms, and notes that "people are most contagious during the early stage of illness" (1). In practice, by the time a child is barking, the household has already been exposed. What is passed around is the virus, not the croup: another child may catch the same parainfluenza and get nothing but a runny nose, while a 2-year-old gets the swollen voice box. If your family seems to be sick constantly, our piece on why you keep getting sick covers what does and does not explain that.

Croup symptoms, in the order they usually arrive

Croup rarely starts as croup. The sequence that clinicians expect is a plain cold first, then the airway symptoms on top of it.

Fever is common but not required. In the AAP's description, most children with viral croup have a low fever, while children with the "spasmodic" pattern — a sudden night-time attack thought to be related to allergy or reflux rather than infection — usually have no fever at all (2). A barking cough with no fever is still perfectly consistent with croup.

Why croup is worse at night

Every source that describes croup says the same thing: symptoms are worse at night (2, 3, 6, 12). The AAP puts the parent's experience bluntly: a child "may go to bed well and wake up in a few hours, gasping for breath" (2).

There is no single published mechanism that explains all of it, and it is worth saying so rather than inventing one. What is well documented is the loop that makes it worse once it starts: being upset and crying worsens the cough and the work of breathing, so a frightened child in a dark room breathes harder, which makes more noise, which frightens everyone further (2). This is also why clinicians assess stridor separately at rest and when agitated — the same child can sound alarming while screaming and quiet two minutes later on a parent's shoulder, and only one of those two states is a severity marker.

The practical consequence for parents is that the daytime version of your child is not a reliable preview of the 2 a.m. version. A pediatrician who saw a cheerful toddler at 4 p.m. is not contradicting you when you call at midnight.

How doctors diagnose croup: by ear and by eye, not by test

This is the part that surprises people, and it is the heart of the condition. There is no swab, no blood test and no scan that makes the diagnosis of croup. It is clinical: the doctor listens, looks and counts.

The American Family Physician 2026 rapid evidence review states it directly — diagnosis is clinical, based on barking cough, hoarseness and inspiratory stridor, with fever sometimes present; radiography or laboratory testing is typically not required and should be considered only if the diagnosis is unclear (4). The Canadian Medical Association Journal review says diagnosis is usually made from history and physical examination alone, with the classic picture being abrupt onset of barky cough, hoarse voice and inspiratory stridor after a day or three of a cold (6).

What the clinician is actually doing in those 90 seconds

• Listening to the cough from across the room. The bark is close to diagnostic on its own.

• Listening for stridor, and noting whether it is there at rest. Stridor only when the child is upset is mild territory; stridor in a calm or sleeping child is not (5, 6).

• Watching the chest and neck. Indrawing — skin sucking in between the ribs, below the ribcage, or at the base of the throat — shows how hard the child is working.

• Watching air move. Decreased air entry is a worse sign than a loud noise. A quieting stridor in a tiring child is a danger sign, not an improvement (5, 6).

• Looking at the child, not the monitor. Alertness, colour, whether the child will drink, whether the child wants to lie down.

Why X-rays are usually not done

A neck X-ray in croup can show a narrowed subglottic airway — the "steeple sign." It sounds decisive and is not. The sign "is neither specific nor sensitive," and it can also appear in epiglottitis, bacterial tracheitis, a tumour or a thermal injury (5). A normal film does not exclude an alternative diagnosis (6).

There is a second, more important reason. Sending a child with a compromised airway off to radiology means upsetting that child, and the CMAJ review warns that imaging "can agitate children and trigger obstruction" when epiglottitis or tracheitis is suspected, so it should be used cautiously and with airway-experienced staff present (6). Radiographs are reserved for atypical symptoms, an unclear diagnosis, or a child who is not responding as expected (6).

So if you take a barking child to an emergency department and leave without a single test, nothing was skipped. That is the standard of care.

The Westley croup score: the five things a clinician counts

When a clinician grades croup, they are usually using — directly or indirectly — the Westley croup score. It is worth knowing what is on it, because it is a tidy list of exactly what matters, and it maps closely onto what you can see at home.

The score comes from a 1978 double-blind trial by Westley, Cotton and Brooks, which needed a way to measure whether nebulized racemic epinephrine changed anything in croup (7). It was built as a research outcome measure and has been used as a severity scale ever since. Five items, 0 to 17 points:

• Level of consciousness — Normal, including asleep: 0 · Disoriented: 5

• Cyanosis (blue colour) — None: 0 · With agitation: 4 · At rest: 5

• Stridor — None: 0 · When agitated: 1 · At rest: 2

• Air entry — Normal: 0 · Decreased: 1 · Markedly decreased: 2

• Retractions (indrawing) — None: 0 · Mild: 1 · Moderate: 2 · Severe: 3

Severity bands: 2 or less is mild; 3 to 7 moderate; 8 to 11 severe; 12 or more is labelled impending respiratory failure (5).

Two things jump out of that table. The heaviest points are not on the noise — they are on consciousness and colour. A drowsy or disoriented child scores 5 on one item alone. And stridor at rest is worth double stridor on agitation, which is the formal version of the rule a parent can use at home.

Canadian guidance describes the same gradient in words rather than points: mild croup is an occasional barky cough with no or limited stridor at rest and little indrawing; moderate is a frequent barky cough with stridor easily audible at rest and visible indrawing but little distress; severe adds prominent stridor, marked indrawing, substantial distress and possible lethargy; and impending respiratory failure looks like a weak cough from fatigue, a quiet stridor, lethargy or reduced consciousness, and blue colour without extra oxygen (6).

Read that last line twice. The most dangerous presentation of croup is quieter than the moderate one.

The dangerous look-alikes, and how croup is told apart from them

This is the real diagnostic work. Croup itself is usually benign; the reason a clinician examines a barking child carefully is to be sure it is croup. The conditions below also cause noisy breathing or airway obstruction in children, and several of them are emergencies.

The one-line versions worth remembering

Croup barks; epiglottitis drools. A child with croup coughs loudly and usually still swallows their own saliva. A child who is drooling, refusing to swallow, leaning forward and frightened, with a high fever and no bark, is a different and far more urgent problem (6, 5).

Epiglottitis is rare now, which is exactly why it gets missed. Haemophilus influenzae type b was the classic cause, and invasive Hib disease has fallen by more than 99% since the conjugate vaccines were licensed; in the early 1980s there were around 20,000 cases a year in the US, mostly in children under 5 (9). It still happens, and it happens disproportionately in children who are not up to date on vaccines — one practical reason the childhood schedule matters; we cover it in the childhood vaccine schedule by age.

Getting better and then suddenly much worse is the tracheitis pattern. Croup should follow a curve: bad night, bad second night, then better. A child who turns the corner and then spikes a high fever and deteriorates needs to be seen again, not reassured that "it's just croup" (6).

No cold beforehand means think again. The foreign body story is a well child who was fine ten minutes ago. No runny nose, no hoarseness, no bark — a sudden choke and then noisy breathing (6).

Minutes, not days, means think allergy. Anaphylaxis can begin within minutes of a trigger, with a tight throat, hoarseness, trouble breathing or swallowing, swelling of the lips or tongue, hives, or pale or bluish colour; in infants it can show as sudden drooling, inconsolable crying or unusual sleepiness. It is treated with epinephrine and a 911 call, not with watchful waiting (10).

Croup season: why this is an autumn article

Croup has a calendar. The 2026 rapid evidence review puts peak incidence in October and November, declining gradually through late April (4). The Canadian review puts the peak in late fall and notes something stranger: odd-numbered years have about 50% more cases, tracking parainfluenza prevalence (6).

CDC's parainfluenza data explain the shape of the year. HPIV-1, the type most associated with croup, produces more cases in the fall. HPIV-2 is also mostly a fall virus and is detected less often. HPIV-3 runs in spring and early summer, and can circulate year-round when the others are out of season. HPIV-4 appears to occur in fall and winter each year (1). A New York surveillance study of 2013 to 2015 found HPIV-4 arriving in late summer and early fall annually, and associated it with more severe illness in children with chronic cardiovascular or respiratory conditions (11).

None of this is preventable with a croup vaccine, because there isn't one. What the autumn calendar does change is the rest of the respiratory picture around it: influenza is the one piece of a child's autumn with a direct, well-evidenced prevention step, which is why the flu shot for kids and when to get a flu shot are October questions. If a shot leaves a child or an adult feeling rough for a day, the normal range of reactions is in our guide to vaccine side effects.

What happens at the appointment, and what doctors do about croup

This article is about recognition rather than treatment, and doses belong to your child's clinician, not to a blog. But it helps to know roughly what the visit looks like, because it tells you what the clinician is weighing.

Corticosteroids are the mainstay. The 2023 Cochrane review pooled 45 randomized trials and 5,888 children and found that, compared with placebo, glucocorticoids may produce greater reductions in croup score at two hours, six hours and twelve hours, all on low-certainty evidence, with very uncertain evidence at 24 hours. The review's authors concluded that the evidence that glucocorticoids reduce croup symptoms at two hours, shorten hospital stays and reduce return visits or readmissions "has not changed in this update." They also note the limits honestly: 98% of the included studies were at high or unclear risk of bias (8).

That is the evidence frame, not an instruction. Which drug, which dose and whether any is needed is a decision for the person examining your child. For moderate to severe croup, nebulized epinephrine may be added, after which children are typically watched for a couple of hours before anyone goes home (4).

The reassuring numbers: population data show 1% to 5% of children with croup are admitted to hospital; of those admitted, fewer than 3% are intubated; and death is rare, estimated at no more than 1 in 30,000 cases (6). In the 2026 review, about 3% of children seen in an emergency department are hospitalized, and up to 5% of those discharged come back (4).

When it is an emergency

Most croup is a loud, horrible, self-limiting two nights. A small share is not, and the signs that separate them are specific.

Call 911 if your child has stridor — the high-pitched noise breathing in — while resting or asleep; is sucking the skin in above, below or between the ribs to breathe; looks blue, grey, pale or blotchy on the lips, tongue or skin; is drooling and cannot swallow; or is limp, unusually sleepy, or not recognizing you. In the UK call 999 (2, 3).

The AAP's own list for when to seek emergency care includes a whistling sound that gets louder with each breath, a child who cannot speak or make sounds for lack of breath, visible struggle to catch breath, bluish lips or fingernails, stridor at rest, and drooling or extreme difficulty swallowing saliva — and says to act on these "even if it is the middle of the night" (2). The NHS adds grunting noises, the tummy sucking in under the ribcage, a child who is unusually quiet and still, limp or floppy, or so distressed and breathless that they cannot be calmed, and a child who is drooling more than usual, struggling to swallow, or wanting to sit rather than lie down. The NHS also advises not to drive to the emergency department yourself (3).

Two quieter signals deserve their own line, because they read as improvement and are not. A stridor that gets softer in a child who is getting more tired, and a child who stops protesting and goes floppy or drowsy, are both severity signs (5, 6). In the Westley score, disorientation alone scores 5 of a possible 17 (5).

Below that threshold, the NHS line is to contact 111 or a GP if you think your child may have croup, and again if the child is getting worse or not improving after being seen (3). Children under 3 months may be referred to hospital as a matter of course (3).

Recurrent croup: when the barking keeps coming back

Some children get croup once. Some get it every autumn, and a few get it several times in a season. Recurrent croup is commonly defined as more than two episodes in a year (12).

The old assumption was that a child who barks repeatedly must have a narrow airway. Current thinking has moved: recurrent croup is now understood as resembling asthma-like airway reactivity more than a fixed anatomical problem, and gastroesophageal reflux disease, asthma and atopy are more common in these children, who usually show no abnormality on bronchoscopy (12). Abnormal findings on airway endoscopy are linked to previous intubation, prematurity and very young age (12).

That does not mean repeated episodes should be shrugged off. The 2026 review advises that recurrent croup should prompt evaluation for anatomic abnormalities or for conditions such as reflux or atopy (4). The practical version for a parent: a dated list of episodes is the thing that turns "he keeps getting croup" into a referral.

Living with a barking cough: the questions people ask next

How long does croup last, and when can my child go back to daycare?

Croup symptoms usually peak around 24 to 48 hours and resolve within about a week, with the barking cough and hoarseness often outlasting the breathing difficulty (12). The NHS says mild croup usually gets better on its own within 48 hours (3). Return-to-daycare rules are set locally and usually hinge on fever and general wellness rather than on the cough itself, because the cough is the slowest thing to go. Your child's doctor or the daycare's policy decides.

Is croup contagious, and can my other children catch it?

The virus is contagious; croup as such is not inherited by whoever catches it. Parainfluenza spreads like other respiratory viruses, with an incubation of generally 2 to 6 days, and people are most contagious in the early stage of illness — before the bark starts (1). A sibling may get the same virus and produce nothing but a runny nose, or may get croup too. Whether your household needs to isolate is a question for your child's doctor.

Can adults get croup?

Adults get the viruses that cause croup, but the classic barking, stridor-producing picture is overwhelmingly a small-child illness because it depends on a small airway — the usual range is roughly 6 months to 3 years, extending to about age 6 (4, 5). An adult with a new hoarse voice and noisy breathing in should be assessed for a different cause, not assumed to have croup.

My child had croup last month and now it's back. Is that normal?

Repeat episodes are common enough to have their own name. More than two episodes a year is the usual definition of recurrent croup, and it is associated with reflux, asthma and atopy more often than with an anatomical narrowing (12). Guidance is to evaluate rather than to keep treating episode by episode (4). Bring a dated list: how many episodes, what month, how long each lasted, whether there was a cold first. Your child's doctor decides whether a referral is warranted.

Does my child need a chest X-ray or a swab to confirm it?

Usually not. Diagnosis is clinical, and testing is typically unnecessary unless the diagnosis is unclear (4). The "steeple sign" on a neck film is neither sensitive nor specific and appears in other conditions too (5). A normal film does not rule out an alternative diagnosis, and imaging a distressed child carries its own risk of making the airway worse (6).

Should I be worried that my child seems fine all day and terrible at night?

That is the ordinary pattern, not a sign that something was missed. Croup is worse at night in essentially every description of it (2, 6, 12). What matters is not the time of day but the signs: stridor at rest, indrawing, colour, and how awake the child is. Call 911 if your child has stridor — the high-pitched noise breathing in — while resting or asleep; is sucking the skin in above, below or between the ribs to breathe; looks blue, grey, pale or blotchy on the lips, tongue or skin; is drooling and cannot swallow; or is limp, unusually sleepy, or not recognizing you.

What about the parent? I have not slept in three nights.

That part is real and is usually left out of croup articles. Several broken nights in a row is a genuine physiological load on the adult doing the carrying, and the 3 a.m. adrenaline does not switch off when the child settles. If you find yourself wide awake and wired after the house is quiet, feeling anxious at night for no reason describes that state and what is actually happening in it. It is not a croup symptom and it is not a character flaw.

Will a humidifier, steam or cold night air fix it?

These are the things every family tries, and the evidence behind them is thinner than their popularity suggests. The 2026 review mentions a single randomized trial in which 30 minutes of outdoor cold air below 50°F, combined with oral dexamethasone, reduced symptoms in mild to moderate croup — one trial, with a drug attached, graded as limited-quality patient-oriented evidence (4). That is not a home protocol. Keeping a child calm, on the other hand, has a clear rationale: being upset and crying worsens the cough and the work of breathing (2). What to do at home for your particular child is your clinician's call.

How to bring this up with your doctor

What to bring. A short, dated log beats a general impression. For each episode: the date, what time the barking started, how long the noisy breathing lasted, whether there was a cold first and for how long, whether there was a fever, whether stridor happened only when crying or also when calm, and what the child looked like — colour, indrawing, how awake. A 20-second phone video of the breathing, taken when the child is calm, is often more useful than any description, because the cough and the stridor are both sounds.

What to say. "She's 2. She had a runny nose for two days, then last night she woke at 1 a.m. with a barking cough and a high-pitched noise breathing in. It lasted about 40 minutes and settled when I held her. No fever. It happened again two nights later, and that time I could hear the noise while she was settled and calm, not just when she was crying." That detail — stridor at rest, not only when upset — is the single most useful piece of information you can give. If you are hearing it while your child rests or sleeps right now, this is not a conversation for an appointment: call 911.

What to ask. "Is this croup, or should we be ruling out anything else?" "Is there stridor at rest right now?" "Which signs mean I call 911 instead of calling you?" "This is her third episode this year — does that change anything?" Ask for the return criteria in plain words and write them down; at 3 a.m. you will not remember them.

If you feel dismissed. The honest reply to "it's just croup" is: "I understand croup is usually mild. What I'd like is to know which specific signs mean this one isn't, and what the plan is if I see them." Asking for the red flags out loud is not challenging the clinician — it is the discharge instruction every guideline expects you to leave with. If the child is getting worse or not improving after being seen, the NHS advice is to come back (3).

And the line that overrides all of this. Call 911 if your child has stridor — the high-pitched noise breathing in — while resting or asleep; is sucking the skin in above, below or between the ribs to breathe; looks blue, grey, pale or blotchy on the lips, tongue or skin; is drooling and cannot swallow; or is limp, unusually sleepy, or not recognizing you. In the UK call 999. Do not drive to the emergency department yourself (2, 3).

Start understanding your body

Learn what affects your energy, stress, sleep, and daily state. Get the app.

Explore your data

How Welltory helps — and what it cannot do

The limits first, and they are unusually important on this topic. Welltory is a general wellness product, not a medical device. It does not diagnose, detect, predict, monitor, prevent, treat or mitigate any condition, including croup. It is built for adults, and it cannot measure anything on your child: Welltory's own help center states that "measurements may not be accurate if you're under 18 or pregnant," because the body systems of people under 18 and of pregnant women work differently and the algorithm is less reliable in those cases (13). So: do not point this app at a barking toddler. It will not tell you whether the airway is narrowing, and nothing in it substitutes for looking at your child.

What it can honestly do is hold the dated log that your doctor will ask for. The question at a croup follow-up is almost always the same — how many episodes, when, how long, what helped — and almost nobody has the answer written down. Welltory is an adult app, so this is you using your own feed as a notebook, not a child tracker.

Note the episode while it is happening. When the app detects a stress or rest stretch on the Today screen, it asks "What happened?" or "What helped you rest?" You can tap a suggested tag, or answer by voice or text in your own words — the app extracts tags such as croup night, up at 3am, ER visit, barking cough — and you can correct them before saving (14). Tag the night of the episode and, if you want the pattern to be useful later, tag what you did about it.

Over a season, My Patterns shows what those tags have in common. It groups your most common stress and recovery tags, flags rare-but-high-impact ones, and for each tag shows frequency, weekday rhythm, monthly trend, heart rate during the episodes, and the full list of episodes. Patterns start to become visible at around seven tagged events, typically needing at least seven occurrences of a tag in the current month plus some history from the previous one (14). Two croup nights in October will not make a pattern. A winter of broken nights will. You can read more in My Patterns.

A record you can actually take to the appointment. The Journal keeps your measurements, tags, personal notes, mood and how you felt physically, day by day (15). From the web app you can export the underlying data as a CSV: log in, open Dashboard, choose a chart, select a time range, and click Export (16); our walkthrough is how to export your health data. What you hand the pediatrician is a dated list of nights — "Oct 14, barking started 1:10 a.m., noisy breathing ~40 min, no fever, settled sitting up" — which is exactly the history that turns a vague story into a decision.

And one honest caveat about what any pattern means. If your own resting heart rate, stress minutes or sleep look worse across a croup week, that is an association with a week of broken sleep and fear, not proof that one caused the other — and it says nothing about your child's airway. A pattern is a connection worth raising with a clinician, not a cause. There is no chart of our own anywhere in this article, and that is deliberate: Welltory has no data on children, because the app is built for adults and its own help center says measurements may not be accurate under 18 (13). A graph assembled from adult users would say nothing about a toddler's airway, so we have not drawn one.

The same red flags apply here, in the same type size. Call 911 if your child has stridor — the high-pitched noise breathing in — while resting or asleep; is sucking the skin in above, below or between the ribs to breathe; looks blue, grey, pale or blotchy on the lips, tongue or skin; is drooling and cannot swallow; or is limp, unusually sleepy, or not recognizing you. No app reading changes that, and no app reading should delay it (2, 3).

How we made it

Made with AI tools, then edited and fact-checked by the Welltory team. See our Editorial & AI policy.

Jane Smorodnikova, co-founder of Welltory, built the methodology behind how we read physiological data.

Written by Tatsiana Yashyna.

Discounts for blog readers: up to 36% off

See what affects your energy, stress, sleep, and daily state with Welltory

Explore your data