Bats are the only mammals that fly. Most navigate and hunt by echo, none are blind, and three species out of 1,400 drink blood — one of which shares its meal with roost-mates who have shared with it.
One in five mammal species is a bat, which is the fact that reorders everything else. They are not a curiosity at the edge of the class; they are a substantial fraction of it, and they got there through two innovations — powered flight and, in most lineages, the ability to build a picture of the world from the echoes of their own voice. Both are more constrained than they sound. Echolocation resolves detail over metres, not the hundreds of metres vision covers, and it forces a trade-off: the calls that carry furthest resolve the least. A bat hunting in open air and the same bat hunting among leaves are using measurably different sensory systems. And the popular version has bats blind, which is wrong twice: echolocating bats see perfectly well, and most Old World fruit bats do not echolocate at all.
Developed record · 75% complete · reviewed 2026-08-09
What this page covers
The order Chiroptera — around 1,400 species, roughly a fifth of mammal species. Split between echolocating bats and the mostly non-echolocating Old World fruit bats.
Quick facts
Species
Around 1,400 — roughly 20% of mammal species
Echolocation
Most bats, but largely not Old World fruit bats
Blood-feeding
3 species out of ~1,400, all in the Americas
Food sharing
Vampire bats regurgitate to roost-mates, kin and non-kin alike
What echolocation actually gives a bat, and what it costs.
Most bats find their way and catch prey by listening to the echoes of their own calls
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Bats of the suborder Yangochiroptera and the horseshoe bat lineage emit ultrasonic pulses and localise objects from the returning echoes. Call design — frequency, bandwidth, duration and repetition rate — varies systematically with foraging habitat, trading detection range against spatial resolution.
Who this applies to
most bats, but explicitly not most Old World fruit bats
Studied in
Chiroptera
You may have heard
“Bats are blind and see with sound”
No bat is blind; many have good vision and some fruit bats rely on it almost entirely. And echolocation is not a substitute for sight — it resolves detail over metres, not the hundreds of metres vision covers, which is why bats navigating long distances use landmarks and the sky rather than echoes.
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Demonstrated by sensory-deprivation experiment in 1941, confirmed by direct recording of emissions, and characterised across families in the decades since. Not seriously disputed.
How far it can be extended
Laryngeal echolocation has been demonstrated across many families independently. Pteropodid fruit bats largely lack it, relying on vision and smell, with the notable exception of the tongue-clicking Rousettus.
Caveats
Most Old World fruit bats do not echolocate; "bats are blind" is wrong twice over, since echolocating bats can also see.
Effective range is tens of metres at most, and only metres for the broadband calls used in clutter.
Echolocation is expensive and is often switched off when other senses suffice.
Still unanswered
How many times did laryngeal echolocation evolve, or was it lost in fruit bats?
How do bats separate their own echoes from those of hundreds of neighbours in a colony?
A prey animal that defeats echolocation by interfering with it, showing the system’s dependence on clean echoes.
The founding experiment is worth knowing because of how cleanly it is built. In 1941 bats were flown through a grid of wires with one sense removed at a time. Blinded, they flew through it. Ears plugged, or mouths covered, they hit the wires repeatedly. Removing one channel at a time is what stops the result being attributed to general impairment — and the hypothesis had been waiting since 1794, unconfirmable until a microphone existed that could hear above 20 kHz.
The trade-off is the part popular accounts leave out. A long, narrow-band, low-frequency call travels far but tells the bat little about shape. A short broadband sweep resolves fine detail and dies within metres. No call does both, so a bat hunting in open air and one hunting inside a hedge are using different sensory instruments — and the same individual switches between them within a single flight.
One tiger moth defends itself by producing clicks that disrupt a bat’s sonar
Well supported
Good evidence backs this, though some details remain open.
The tiger moth Bertholdia trigona produces high-rate ultrasonic clicks when attacked. Naive big brown bats capture silenced moths readily but repeatedly fail against clicking ones, and — unlike responses to warning or startle signals — do not improve with experience over successive nights.
Who this applies to
one tiger moth species tested against one bat species
Studied in
Bertholdia trigona
Why we rate it this way, and what the caveats are
Well supportedModerate confidence
The learning-curve design distinguishes jamming from the two competing explanations cleanly. Confidence is moderate rather than high because it rests on one moth species and one bat species in a flight room.
Caveats
Many tiger moths click; most are thought to be warning of toxicity rather than jamming.
One bat species, in captivity, naive to the moth.
The precise acoustic mechanism of the interference is not fully characterised.
Still unanswered
How many of the roughly 11,000 tiger moth species jam rather than warn?
Bats failed against clicking moths and never learned to overcome them, ruling out aposematic warning and startle.
How we know
The moth that bats never learned to catch
A tiger moth clicks when a bat attacks. Is it warning that it tastes bad, startling the bat, or interfering with the sonar itself?
The three hypotheses are hard to separate from a single encounter and easy to separate over many, because they predict different learning curves. A warning is learned — a bat should stop attacking once it associates the click with a bad meal. A startle habituates — the bat should stop being surprised. Jamming keeps working regardless. Naive big brown bats were flown against tiger moths that were either able to click or had been silenced, over successive nights, with every encounter recorded acoustically and on high-speed video.
What happened
Bats captured silenced moths readily. Against clicking moths they failed repeatedly — and, over successive nights, never improved.
What it shows
The clicks interfere with echolocation rather than warning or startling. The flat learning curve is the discriminating evidence: a bat can learn around a warning and habituate to a startle, but it cannot learn its way out of a jammed signal.
What it does not show
One moth species against one bat species in a flight room. Most clicking tiger moths — of roughly 11,000 species — are thought to be warning of toxicity rather than jamming, so this is a demonstration that jamming exists, not that it is common.
The controls — what makes this evidence rather than a story
Silenced moths of the same species isolated the clicks as the variable, holding taste, appearance and flight behaviour constant.
Bats naive to the moth ensured no prior learning.
Repeating encounters across nights is the actual experiment: the shape of the curve discriminates the hypotheses.
The jamming result is a nice piece of experimental reasoning. Clicks could warn of toxicity, startle, or interfere. Those predict different learning curves: a warning is learned, a startle habituates, and jamming keeps working. The bats never adapted.
Words used here
Echolocation
Building a picture of surroundings from the echoes of self-produced sound.
Bandwidth
The range of frequencies in a call. Wide bandwidth resolves detail; narrow bandwidth carries further.
Vampire bats share blood with roost-mates. Is that kinship, or is it repayment — and can the two be told apart?
In the wild they cannot: bats that roost together are usually related, so relatedness and association always travel together. The captive colony solved it by doing what the field cannot — choosing who goes hungry. Individual bats of known pedigree were fasted in turn and returned to the group, and every donation was recorded alongside genetic relatedness, the history of who had previously given to whom, and how much the hungry bat harassed potential donors.
What happened
How much a bat had previously donated to a given partner predicted how much it received from that partner far better than genetic relatedness did, and better than harassment.
What it shows
Reciprocity, not kin selection, dominates blood sharing in this species. It also shows how a contested field observation gets settled — the 1984 study was not wrong, merely unable to separate the explanations.
What it does not show
Reciprocity here need not involve any bookkeeping by the bats; a simple bond-strength rule reproduces the same pattern without anyone keeping score. Twenty animals in a captive colony, with donations concentrated among a few strong partnerships, and no predation or roost dynamics.
The controls — what makes this evidence rather than a story
Fasting was assigned by the experimenters rather than arising naturally, breaking the usual correlation between need and circumstance.
Known pedigree allows relatedness to be measured rather than inferred from association.
Harassment was recorded as a competing explanation, not assumed away.
A vampire bat that has fed will feed a hungry roost-mate — especially one that once fed it
Well supported
Good evidence backs this, though some details remain open.
Common vampire bats regurgitate blood to roost-mates including non-kin. In controlled fasting trials where relatedness, prior donation and harassment are measured simultaneously, a partner’s history of giving predicts donations substantially better than genetic relatedness does.
Who this applies to
common vampire bats, in wild roosts and captive colonies
Studied in
Desmodus rotundus
Why we rate it this way, and what the caveats are
Well supportedHigh confidence
A field observation contested for nearly thirty years, then settled by a controlled design that could manipulate who went hungry and separate relatedness from association.
Caveats
Reciprocity here need not involve any tallying by the bats; a simple bond-strength rule reproduces the pattern.
The decisive evidence is from a captive colony.
Vampire bats face an unusually severe starvation deadline, which may be why this evolved here.
Still unanswered
Do bats track individual partners, or simply invest in whoever they groom most?
How are cheats — bats that take without giving — handled over long timescales?
The original wild observation, suggestive but unable to separate kinship from association.
Vampire bats are the standard case in the reciprocity literature for a practical reason: the stakes are measurable. A bat that misses two consecutive nights of feeding dies, so a donation is genuinely costly and a repayment genuinely valuable. That makes the accounting visible in a way it rarely is.
The story also shows how a contested result gets settled. The 1984 field observation could not separate kinship from association — bats that roost together are often related. Thirty years later, a captive colony of known pedigree allowed the experimenters to choose who went hungry, and prior giving turned out to predict receiving better than relatedness did.
Three of roughly 1,400 bat species feed on blood, all in the Americas. The rest eat insects, fruit, nectar, fish or frogs.
Words used here
Reciprocal altruism
Helping someone at a cost to yourself, on the expectation of being helped in return later.
A colony of fruit bats sounds like undifferentiated squabbling. Recorded around the clock for months and fed to a classifier, the squabbling turns out to encode who is calling, who is being addressed, and what about — food, a mating attempt, a disputed sleeping position.
The careful reading matters here. Information being recoverable by an algorithm is not the same as information being used by a bat. It raises the ceiling on what might be going on without demonstrating that any receiver extracts it, and playback experiments are what would settle that.
A bat hangs because it cannot take off from the ground the way a bird can. Bird wings generate lift from a standing start against strong legs; a bat’s wings are hands, its legs are light and rotated for hanging, and the whole animal is built to drop into flight rather than leap into it. Hanging from a high place is a launch position.
It costs almost nothing to hold. The tendons in a bat’s feet lock closed under the animal’s own weight, so hanging requires no muscular effort at all — a dead bat stays hanging. Releasing the grip is the part that takes work.
Roosts range from caves and hollow trees to bridges, roof spaces and, in some tropical species, tents cut from leaves.
Most temperate bats roost socially, which conserves heat; some species roost alone.
Torpor during the day is common, letting a bat drop its body temperature and save energy between nights.
Hibernation in temperate species is a separate, deeper state lasting weeks, and rousing from it is expensive.
That last distinction matters more than it sounds. A hibernating bat disturbed in winter burns fat reserves it cannot replace before spring, which is why cave disturbance is a conservation issue rather than a nuisance.
Bats are the only mammals capable of powered flight. Flying squirrels and colugos glide, which is a different thing — they descend, gracefully.
Words used here
Torpor
A short-term drop in body temperature and metabolism, often daily. Deeper and longer, it becomes hibernation.
Patagium
The membrane forming a bat’s wing, stretched between elongated finger bones, the body and often the legs and tail.
A mammal the size of a mouse living the reproductive life of an animal thirty times its weight.
Bats are mammals and everything follows from that: no eggs, live birth, milk. What is unusual is the pace. A mouse of similar size produces several litters a year of half a dozen young each and is dead within two. Most bats produce a single pup a year, and a common pipistrelle weighing five grams can live past sixteen. For their body size, bats are among the longest-lived mammals known — a Brandt’s bat recaptured in Siberia was over forty years old at a weight of about seven grams. Slow reproduction and long life go together, and they are why a bat population that crashes takes decades to recover.
Mating in temperate species happens in autumn, before hibernation, and then the timing gets strange. The female stores the sperm through the winter — viable, unused, for months — and ovulates in spring when she emerges. Fertilisation happens then. This is not a quirk of a single species: sperm storage over winter is widespread in temperate bats, and it solves an obvious problem. Mating requires two bats to find each other and there is no better moment for that than the autumn swarming, when hundreds gather at cave entrances. Pregnancy requires food, and there is none until spring.
Gestation runs six to nine weeks once it starts, varying with how cold the spring is — a cold spring slows the pregnancy itself.
Females gather in maternity roosts, sometimes thousands strong, and males are largely absent.
Birth is usually feet first, which is unusual among mammals and sensible for an animal giving birth while hanging.
The pup is enormous relative to the mother — up to a quarter of her weight — and is born furless and unable to fly.
First flight comes at three to five weeks; weaning follows shortly after.
In a maternity roost of several thousand, a returning mother finds her own pup among all the others. She does it by call and by smell together, and she is right the great majority of the time. Given that the pups are packed at densities of hundreds per square metre and are all calling at once, that is a substantial recognition problem, solved twice a night for weeks.
Bats do not lay eggs, and no mammal does except the platypus and the echidnas. The question is common enough to be worth answering directly: a bat is not a bird, and the wing is a hand.
A bat wing is a hand with the fingers enormously extended and a membrane stretched between them. That single fact separates bat flight from bird flight completely. A bird’s wing is a rigid aerofoil with feathers; a bat’s is a flexible sheet whose shape the animal changes continuously, joint by joint, through the wingbeat. It is less efficient over distance and far more manoeuvrable, which is why a bat can turn inside its own body length to take an insect and a swift cannot.
The membrane is living tissue, laced with blood vessels and touch receptors, and it heals from tears. It also has to be kept supple, which is what a bat is doing when it appears to be grooming — the oils come from glands at the muzzle and are worked across the wing.
Taking off from the ground is the well-known limitation and it is often overstated. Most bats cannot launch from a flat surface: the legs are light, rotated for hanging, and the wings need airspace beneath them. Several species can, though — vampire bats spend time on the ground near their prey and launch with a powerful jump from the forelimbs, which is a genuine exception rather than a technicality. A bat found on the floor is usually in trouble, and should not be picked up with bare hands.
Bats swim, and reasonably well, using the wings as paddles. It is an emergency capability rather than a habit — a bat that lands on water can usually get to an edge and climb out.
Why the night shift at all is a better question than it looks, and the answer is not that bats hate light. Two explanations carry most of the weight. Insect-eating bats hunt a night-flying insect fauna that is enormous and largely uncontested. And a bat flying by day would be exposed to birds of prey, against which a slow-turning mammal with no speed has few options — which is supported by what happens on islands with no raptors, where some bats do fly in daylight. Bats are also not blind: every bat can see, and the large fruit bats have excellent vision and no echolocation at all.
A bat circling a street light is hunting, not confused. Lights concentrate moths, and some bat species have learned to exploit that so reliably that lighting changes measurably shift where they forage — while other, slower-flying species avoid lit areas entirely, which is why lighting is a conservation issue.
Words used here
Patagium
The wing membrane, stretched between elongated finger bones, the body and often the legs and tail.
Nocturnal
Active at night. For bats this is near-universal and appears to be driven by prey availability and by predators rather than by light itself.
A bat in a roof space is usually a maternity roost, and it is usually not new. Buildings replaced the hollow trees and caves that temperate bats used before there were buildings, and a roost site is used for decades — often by the same lineage of females returning each summer. A single bat that appears inside a living room in late summer is normally a juvenile that has flown down a gap rather than out of one, and is lost rather than nesting.
Bats do not chew wiring or timber, build nests or bring in material. What they leave is droppings, which are dry, crumble to powder and consist almost entirely of insect fragments.
Safety
If a bat is in the room
Open a window, close the internal doors, turn the lights down and leave it to find its own way out. If it must be moved, do not handle it with bare hands — use gloves and a box, and contact a bat helpline or local wildlife service. Anyone bitten or scratched by a bat should seek medical advice promptly, because post-exposure treatment for rabies is highly effective and depends on being started early.
Where this applies: Bats and their roosts are protected by law across the UK, the EU and many other jurisdictions, and disturbing a roost is generally an offence. Rabies risk from bats varies by region.
When to get help: Contact a national bat helpline or local wildlife authority before any work affecting a roost, and a medical service immediately after any bite or scratch.
The rabies question is covered in its own section, and the short version belongs here too: the risk from an undisturbed bat is very low, and it is not zero if you handle one. Both halves of that sentence matter.
The great majority of the roughly 1,400 bat species eat insects, and they eat a great many: a single small insectivorous bat can take hundreds of insects in a night, and colonies measured in millions remove tonnes of them. This is the group’s main ecological role, and the main reason bat decline is an agricultural problem as well as a conservation one.
Insects: most species, caught on the wing by echolocation, or gleaned from surfaces.
Fruit and nectar: the large fruit bats, which pollinate and disperse seeds across the tropics — durian, agave and many forest trees depend on them.
Fish, frogs and small vertebrates: a small number of specialists.
Blood: three species, all in the Americas, and only one of which regularly feeds on mammals.
The three vampire species get a disproportionate share of the attention and are worth keeping in proportion: three out of about fourteen hundred. They are also the group that produced one of the more remarkable results in animal behaviour, sharing blood meals with roost-mates that failed to feed.
The most consequential question people ask about bats. NatureHQ answers the shape of it and sends you to a clinician for the rest.
Direct contact with a bat is treated as a possible rabies exposure, and the response is urgent — because a bat bite can be too small to notice
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Rabies is almost invariably fatal once clinical signs appear and is almost always preventable by prompt post-exposure prophylaxis. In regions where dog-mediated rabies is controlled, bats are the principal wildlife reservoir. Bat teeth are fine enough that a bite may leave no visible wound, so public health guidance treats direct bat contact — including waking to a bat in a room — as a possible exposure requiring same-day medical assessment rather than requiring evidence of a bite.
Who this applies to
bats in regions where rabies or related lyssaviruses are present; the assessment principle is international, the protocol is nationalDo not extend this beyond the taxa listed — the popular version over-reaches.
Studied in
Chiroptera
You may have heard
“Bats are riddled with rabies”
Prevalence in wild bat populations is low, and most people who see a bat are at no risk whatsoever. What makes the guidance strict is not how likely infection is but how bad it is if it happens, combined with a bite that can be invisible. Treating bats as diseased vermin has driven roost destruction, which scatters bats into buildings and increases contact — the opposite of what it intends.
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
International expert consensus on rabies, consistent across national public health agencies for decades. The near-total fatality of untreated symptomatic rabies and the near-total effectiveness of prompt prophylaxis are not in dispute.
How far it can be extended
Rabies status varies enormously by country and region, and a small number of territories are rabies-free. Whether this applies to a given encounter is a question for a local health authority, not for a nature reference.
Caveats
NatureHQ states that assessment is urgent. It does not state what treatment is indicated; that is a clinical decision and it varies by country.
Rabies status differs by region and changes; some territories are rabies-free and the assessment there differs.
The vast majority of bats are not infected. The reason for caution is the severity of the outcome, not the likelihood.
Handling bats is restricted or prohibited in many jurisdictions, and most species are legally protected.
Still unanswered
How often do unrecognised bat bites actually occur relative to recognised ones?
How much does lyssavirus prevalence vary between bat species and regions?
O'Shea et al., 2014 · Emerging Infectious Diseases
Context on why bats host a notable viral diversity — and a caution against the roost destruction that framing invites.
Two facts drive all the guidance. Rabies is essentially always fatal once symptoms appear and essentially always preventable if treated promptly — so the window matters more than almost anything in medicine. And a bat’s teeth are fine enough that a bite can leave no mark you would notice, which is why public health bodies do not ask whether you were bitten.
Safety
If you have had direct contact with a bat
Contact a doctor or your national health service the same day — not next week, and not only if a wound appears. This includes handling a bat with bare hands, being scratched or bitten, and waking to find a bat in the room with you or with a child or someone who cannot report a bite. Wash any wound thoroughly with soap and running water while you arrange the call. Do not handle the bat to bring it in for testing; if it needs to be contained, your health authority or a licensed bat handler will tell you how. Whether treatment is needed, and which, is a clinical decision that depends on your country, the species and the circumstances — NatureHQ does not make it and no website should.
Where this applies: International principle; the actual protocol is set by your national or regional health authority and differs between countries.
When to get help: A doctor or national health service, the same day, after any direct contact with a bat in a region where rabies or a related lyssavirus occurs.
Set against that, the risk of encountering an infected bat is low: prevalence in wild populations is small, and most people who see a bat are at no risk at all. The guidance is strict because of how bad the outcome is, not because infection is likely.
The opposite error does real harm too. Bats do host a notable diversity of viruses — one proposal links this to an immune system shaped by the extreme metabolic demands of flight — and the conclusion people draw, that bats should be cleared out of an area, backfires. Destroying a roost scatters bats into buildings and increases human contact. Most bat species are also legally protected, and disturbing a roost is an offence in much of Europe and North America.
Practical
A bat in the house, with nobody in contact with it
If a bat is flying in a room and nobody has touched it, the usual advice is to close the door to the rest of the house, open a window wide, turn the lights down and leave it to find its own way out. Do not swat it or catch it bare-handed. If it does not leave, or if it is grounded, contact a bat conservation organisation or your local wildlife authority — in many countries there are volunteer bat carers who do exactly this.
Where this applies: General practice; check your national bat conservation body for local rules.
Words used here
Lyssavirus
The genus of viruses that includes classical rabies and the bat-associated viruses related to it.
Post-exposure prophylaxis
Treatment given after a possible exposure and before symptoms, which is what makes rabies preventable. It is a clinical decision, not a self-assessment.
Lazzaro Spallanzani found that blinded bats flew and navigated normally while deafened ones could not, and concluded that bats somehow orient by hearing. Nobody could say how, and the finding was largely dismissed for a century and a half.
1941
Landmark experiment
Echolocation demonstrated and named
Griffin and Galambos flew bats through wire arrays with vision, hearing or the mouth blocked in turn, and — using a newly available ultrasonic detector — recorded the emissions directly. The pulse rate rose as obstacles approached.
Changes how the 1794 result reads
Spallanzani was right and unprovable. What changed was not the reasoning but the instrument: no microphone existed that could hear above 20 kHz until the 1930s, so the mechanism could not be observed at all.
Marked bats in Costa Rican roosts were seen regurgitating blood to non-kin, with association the strongest predictor of a donation — suggestive of reciprocity, and unable to separate it from kinship.
A synthesis across insectivorous families showed echolocation call structure tracking the perceptual task set by habitat, with the same bat shifting design as it moved between open air and clutter.
Changes how the 1941 result reads
Echolocation stopped being a single fixed ability a species either has or lacks, and became a set of compromises between range and resolution that an individual bat selects between from moment to moment.
Big brown bats failed repeatedly against clicking tiger moths and never learned to overcome them — a learning curve inconsistent with warning or startle, and consistent with interference.
Reciprocity in vampire bats confirmed under control
Controlled fasting in a colony of known pedigree allowed relatedness, prior giving and harassment to be measured at once. Prior giving predicted receiving substantially better than relatedness did.
Changes how the 1984 result reads
The field study was not wrong, merely unable to separate the explanations — bats that roost together are usually related. Choosing who goes hungry is the manipulation the wild does not permit, and it settled a question that had been open for twenty-nine years.
How many times did laryngeal echolocation evolve — once with a loss in fruit bats, or several times independently?
Why it matters: The answer changes how remarkable echolocation is: a single origin makes it a one-off innovation, several makes it something bat bodies are predisposed to arrive at.
How does a bat separate its own echoes from those of hundreds of neighbours in a crowded roost?
Why it matters: It is a signal-processing problem engineers find hard, solved routinely by an animal leaving a cave in a column of thousands.
How many clicking tiger moths jam, and how many merely warn?
Why it matters: Jamming is demonstrated in one species out of roughly 11,000, so whether it is an oddity or a widespread strategy is entirely open.
Do bats use the addressee information their calls appear to contain?
Why it matters: A classifier recovering information from a recording is not a bat using it, and the gap between those two things is where most overstatement about animal communication happens.