How Many Neurons Are in the Human Brain?
The human brain contains roughly 86 billion neurons. Learn about these cells, how they connect, and why the number matters.
The human brain has about 86 billion neurons. Not 100 billion. The 100 billion figure was repeated in textbooks for decades, but nobody ever actually counted that many. The real number comes from a 2009 study by Brazilian neuroscientist Suzana Herculano-Houzel and her team, who developed a method that involved, essentially, dissolving brains into soup and counting the nuclei.
That study also killed a second myth: the idea that glial cells outnumber neurons ten to one. They don't. The ratio is closer to 1:1.
And here's the part that still gets me even after reading through dozens of papers on this: about 80% of your neurons are in the cerebellum, a structure that makes up only 10% of total brain mass. The cerebral cortex, the part most people think of as "the brain," has about 16 billion. That 16 billion is probably the number that matters most for human cognition, because no other land animal has more cortical neurons than we do. But I'll get to that.
The human brain also isn't special in its design. It's a primate brain, built according to the same scaling rules as a marmoset's. It's just bigger. That's less poetic than what most people expect, but it's what the data shows.
Where the "100 billion" came from
In 1895, the first published estimate of total brain neurons was about 3 billion. Herbert Haug estimated 70–80 billion in 1986. Williams and Herrup, in a well-cited 1988 Annual Review of Neuroscience paper, compiled partial counts from the literature and arrived at roughly 85 billion. Close to what we now think is correct.
So where did 100 billion come from? It seems to have appeared around 1991 as a convenient round number and then spread through citation chains where textbook authors cited each other instead of primary data. Eric Kandel's Principles of Neural Science stated it in both the third (1991) and fourth (2000) editions. When Herculano-Houzel asked Kandel for the original source, he couldn't name one.
Roberto Lent, a Brazilian neuroscientist, actually titled his Portuguese-language textbook One Hundred Billion Neurons. He later added a question mark.
Von Bartheld, Bahney, and Herculano-Houzel traced the full history of brain cell counting in a 2016 review, "The Search for True Numbers of Neurons and Glial Cells in the Human Brain" (Journal of Comparative Neurology, 524(18):3865–3895). Published estimates over 150 years ranged from 3 billion to 1 trillion. The 100 billion number crystallised as consensus around 1991 without anyone doing the actual work to verify it. That's an astonishing thing to have happened in science, and yet here we are.
Dissolving the brain into soup
Herculano-Houzel is now at Vanderbilt University. Before that, she was doing science communication at a museum in Rio de Janeiro. Someone asked her how many neurons the human brain has. She looked it up. Nobody actually knew.
The method
In 2005, she and Roberto Lent published the isotropic fractionator in the Journal of Neuroscience (25(10):2518–2521). The idea is almost comically direct. You take brain tissue preserved in paraformaldehyde, cut it into defined regions, and homogenise it in a detergent (Triton X-100) that dissolves cell membranes but leaves nuclear membranes intact. You end up with a suspension of free-floating cell nuclei. Brain soup.
To get a total cell count, you stain the nuclei with DAPI (a fluorescent DNA dye), count samples under a microscope with a haemocytometer, and scale up to the total volume. To figure out which nuclei came from neurons versus other cells, you label a separate sample with anti-NeuN antibody, which binds a protein found only in neuron nuclei. The percentage of NeuN-positive nuclei tells you the neuron fraction.
It takes about 24 hours versus weeks for stereology. It's cheap. It works. The trade-off is that you lose all spatial information: you know how many neurons a region has but not how they're arranged.
The result
The big human brain paper came out in 2009: Azevedo et al., "Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain" (Journal of Comparative Neurology, 513(5):532–541). Four adult male brains, ages 50 to 71. The count: 86.1 ± 8.1 billion neurons and 84.6 ± 9.8 billion non-neuronal cells. Individual brains ranged from 78.8 to 95.4 billion neurons.
The paper was rejected by Nature, PNAS, Neuron, and the Journal of Neuroscience before it landed in the Journal of Comparative Neurology. It now has over 3,000 citations. These things happen.
A 2013 follow-up by Andrade-Moraes and colleagues looked at five older female brains and found a lower average, about 67 billion (range: 61–73 billion). That's a meaningful drop, and it hints at sex and age effects that four male brains couldn't capture.
The 2025 challenge
In March 2025, mathematician Alain Goriely published a paper in Brain (148(3):689) asking whether a sample of four brains is enough to pin down the number. His argument: the data supports a range of 61–99 billion, not a precise point estimate. Roberto Lent pushed back in Brain (148(5):e37), noting that 86 billion converges with earlier stereological estimates of 70–85 billion. Herculano-Houzel also responded (148(8):e72), arguing that biological variation is the point: different people genuinely have different numbers of neurons.
The honest answer, if there is one, is that the human brain has roughly 86 billion neurons, give or take, with a realistic range of maybe 60 to 100 billion depending on who you are, how old you are, and how you measure.
How scientists count neurons
The basic problem is deceptive: the brain is a three-dimensional structure, and you're looking at it in thin two-dimensional slices. Every counting method is, at some level, trying to deal with that.
Profile counting (1800s–1970s)
Early researchers cut thin sections (5–15 µm), stained them, and counted the cell bodies they could see. These are profiles, not cells. A big neuron shows up in more slices than a small one, so it gets overcounted. Correction factors (Abercrombie, 1946, and others) tried to compensate, but results swung wildly depending on assumptions about cell size, section thickness, and tissue shrinkage. Two orders of magnitude of variation. Not great.
Design-based stereology (1980s–present)
Sterio introduced the disector in 1984, which uses pairs of sections to count cells in 3D without assumptions about their shape. The optical fractionator, from West, Slomianka, and Gundersen (1991) (Anatomical Record, 231(4):482–497), combined this with systematic random sampling and became the standard.
Bente Pakkenberg at Bispebjerg University Hospital in Denmark applied it to human brains and produced some of the best data we have on cortical neuron numbers, sex differences, and aging. But stereology is painfully slow. A single mouse brain can take a week. Doing the whole human brain comprehensively was impractical before the soup method came along.
Validation
Does the brain soup method give the same answers as stereology? Bahney and von Bartheld (2014) compared them directly (Journal of Neuroscience Methods, 222:165–174). No significant difference (p > 0.4). Ngwenya et al. (2017) confirmed this in another species. The methods agree.
New techniques (2020s)
The latest approaches don't replace the isotropic fractionator for total counts, but they tell you what the neurons actually are. Single-cell RNA sequencing reveals the molecular identity of each cell. The BRAIN Initiative Cell Census Network (BICCN) used it to identify 3,313 transcriptomic cell subclusters in the adult human brain. Spatial transcriptomics (MERFISH, Slide-seq) maps gene expression in place. Google and Harvard reconstructed 1 mm³ of human temporal cortex at nanometre resolution: 57,000 cells, 150 million synapses, 1.4 petabytes of imaging data (Shapson-Coe et al., Science 384, 2024).
Where the neurons are
This is the part that surprises people.
The cerebellum: 69 billion in a fist-sized structure
The cerebellum is small, about 10% of total brain mass. It has roughly 69 billion neurons, about four-fifths of the total. Almost all of these are granule cells, tiny neurons packed at enormous density. They need very little glial support (the glia-to-neuron ratio in the cerebellum is only 0.05–0.23).
It handles motor coordination and balance. It's also increasingly implicated in language, cognition, and emotional regulation, though the details are still being worked out. About 20 million nerve fibres connect it to the cerebral cortex on each side.
The cerebral cortex: 16 billion
The cortex has about 19% of total neurons, but its 16 billion are the biggest and most connected. Each cortical pyramidal neuron can receive thousands of synaptic inputs. Total synapse count across the neocortex: about 150 trillion. No other land animal's cortex comes close to 16 billion neurons.
Earlier estimates ranged from 10 to 26 billion depending on method and whose brains were counted. Pakkenberg and Gundersen (1997) reported about 23 billion for males and 19 billion for females.
Everything else
The thalamus, hypothalamus, basal ganglia, and brainstem together have only about 700 million neurons, despite being involved in consciousness, arousal, autonomic regulation, and motor control. Glia outnumber neurons about 10:1 in these regions because the neurons are large and metabolically expensive. The spinal cord adds another 200 million or so.
The 10-to-1 glia myth
The claim that glial cells outnumber neurons 10 to 1 goes back to Holger Hydén at the University of Gothenburg in 1960. He was probably looking at brainstem vestibular nuclei, where glia really do outnumber neurons by roughly that much. But he stated it as a general fact, and everyone repeated it: Kuffler and Nicholls in From Neuron to Brain (1976), Kandel and Schwartz in Principles of Neural Science (1981 onward), Bear, Connors, and Paradiso in Exploring the Brain (2001).
The 2009 count found 84.6 billion non-neuronal cells alongside 86 billion neurons. Close to 1:1. Subtract endothelial cells (20–25 billion cells lining blood vessels) and actual glia number about 40–50 billion, which might be fewer than neurons. Oligodendrocytes are the biggest fraction (45–75%), then astrocytes (19–40%), then microglia (about 10%).
The cerebellum is why the overall ratio comes out near 1:1. Its 69 billion granule cells barely need any glial support. Remove the cerebellum and the rest of the brain has a glia-to-neuron ratio closer to 4:1.
How the number changes across people and lifetimes
Sex differences
Pakkenberg and Gundersen's 1997 study of 94 Danish brains found that males have about 23 billion neocortical neurons on average, females about 19 billion. That's a 16% difference. Males also had about 28% more neocortical glial cells. But the range across all healthy adults is huge, 14.7 to 32.0 billion, with plenty of overlap between sexes.
That 16% gap doesn't correspond to any comparable gap in IQ. This is one of the cleaner pieces of evidence that the wiring matters more than the count, at least within our species.
Aging
The popular image of aging as mass neuron death is mostly wrong. Between age 20 and 90, total neocortical neuron loss is about 10%, roughly 85,000 per day, about one per second (Pakkenberg et al., 2003, Experimental Gerontology 38:95–99). Not trivial, but nothing like the catastrophe people imagine. White matter degradation and synaptic changes probably do more damage to cognition with age than losing neurons does.
Neurodegeneration
Alzheimer's disease is a different story entirely. The hippocampal CA1 region, which barely loses any neurons in normal aging, degenerates severely. The entorhinal cortex goes early; widespread cortical loss follows. In Parkinson's, 60–80% of dopaminergic neurons in the substantia nigra are already gone by the time motor symptoms show up.
Adult neurogenesis
Whether adult humans grow new neurons is still unresolved. Some evidence (BrdU labelling, carbon-14 dating, single-cell RNA-seq) supports low-level neurogenesis in the hippocampus. A 2018 Nature study (Sorrells et al.) found almost none in adults. A 2025 review by Simard et al. in The Neuroscientist (31:141–158) suggests it probably happens, but slowly, and might be therapeutically targetable. The field hasn't settled this.
Comparative perspective
The human brain isn't the biggest, and it doesn't have the most neurons. What it does have is the most cortical neurons of any land animal.
Elephants
African elephants have about 257 billion total neurons. Three times the human count. But 97.5% of them (about 251 billion) are in the cerebellum, probably for controlling the trunk, which weighs about 100 kilograms and can pick up a single blade of grass. The elephant cortex has only about 5.6 billion neurons, roughly a third of the human total.
If you want a clean demonstration that cortical neurons, not total neurons, track with cognitive ability across species, this is it.
Primate scaling
Herculano-Houzel's most interesting finding might not be the 86 billion itself. It might be the scaling rules, described in her 2012 PNAS paper.
In primates, ten times more neurons means roughly ten times more brain mass. In rodents, it means 35–50 times more brain mass, because rodent neurons get physically larger as brains scale up. Primates don't do this. They keep neurons small and pack more of them in. A capybara and a bonnet monkey have similar-sized cortices (about 48 grams each), but the monkey has 5.5 times more cortical neurons: 1.7 billion versus 306 million.
Humans are right where you'd expect on the primate curve. Our brain has the number of neurons predicted for a primate of our body size. Nothing structurally unique about it. Herculano-Houzel's argument is that cooking, invented roughly 1.8 million years ago, was what made it affordable. The brain eats about 20 watts at rest, roughly 20% of the body's energy budget. You need a lot of calories to run 86 billion neurons, and cooking made food energy-dense enough to cover the bill.
Other animals
Chimps have about 28 billion neurons (7 billion cortical). Gorillas about 33 billion (9 billion cortical). Dogs 2.25 billion. Cats 760 million. Mice 71 million. An octopus manages 500 million across a body plan with nothing resembling a cortex. Honeybees navigate with about 960,000. The fruit fly, whose complete connectome was mapped in 2024 by the FlyWire project, runs on 140,000–160,000 neurons. C. elegans has 302.
I keep thinking about the fruit fly number. 140,000 neurons. 50 million synapses. And it can fly, navigate, learn, court mates, fight, and sleep. If that doesn't make you reconsider what "intelligence" requires, I don't know what would.
When neuron count predicts things and when it doesn't
Across species
Cortical (or pallial, in birds) neuron count is the best single predictor of cognitive ability across species. Corvids and parrots, which have the most pallial neurons among birds, match primates on many cognitive tests. The old metric, encephalisation quotient (brain-to-body mass ratio), doesn't work well because it treats a gram of rodent brain the same as a gram of primate brain.
Within humans
Inside our species, neuron count stops predicting much. Pakkenberg et al. (2020) looked at 50 male brains and found no correlation between neocortical neuron number and IQ (Cerebral Cortex 31:650–657).
What does predict IQ, at least partly? Goriounova et al. (2018) found that people with higher IQ scores had larger, faster-conducting pyramidal neurons with more complex dendritic trees. That accounted for about 25% of the variance. Synaptic density and network connectivity seem to matter too. It's about the kind of neurons you have and how they're wired, not how many.
Brain cell atlases
2023 to 2025 has been a data flood.
In October 2023, the BICCN published 24 papers across Science and related journals, including the first draft Human Brain Cell Atlas v1.0. Over 3 million cell nuclei from about 100 dissections, 3,313 cell subclusters organised into 461 clusters and 31 superclusters (Siletti et al., Science 382). About 80% of the identified types were neuronal. The brainstem turned out to have the most cell type diversity, which wasn't necessarily expected.
Companion papers mapped the developing human brain from 5–14 postconceptional weeks. Two December 2023 Nature papers gave us a complete mouse brain cell atlas: 5,322 transcriptomic clusters from about 7 million cells. A 2024 atlas in Nature Medicine pulled together 26.3 million cells from 70 human and 103 mouse studies.
BICAN (BRAIN Initiative Cell Atlas Network) has about $500 million over five years to build reference atlases across the full human lifespan. A 2025 Nature collection added developmental atlases.
None of this replaces the isotropic fractionator for total counts. Single-cell transcriptomics tells you what kinds of cells exist, not how many of each there are. But the two approaches work together, and it's becoming obvious that "86 billion neurons" doesn't begin to capture the complexity. There are thousands of molecularly distinct neuron types, and cataloguing them is going to take years.
Wrapping up
The human brain has about 86 billion neurons. That's an approximate number from a small sample. With data from other studies folded in, a realistic range is probably 60–100 billion, varying by age, sex, and the specific person.
The distribution is lopsided. Four-fifths of neurons are in the cerebellum. The 16 billion cortical neurons seem to matter most for what we think of as human cognition. Neurons and glia are in roughly equal numbers, not 10:1. And the brain's architecture is just a scaled-up primate design. No special sauce.
We have the count now, more or less. What those 86 billion neurons actually do, how they wire up and how they fail in disease, is less clear. The atlas projects are making progress, but there's a lot of brain left to catalogue.
References
Azevedo, F. A. C. et al. (2009). Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. Journal of Comparative Neurology, 513(5), 532–541. DOI: 10.1002/cne.21974
Herculano-Houzel, S. & Lent, R. (2005). Isotropic fractionator: a simple, rapid method for the quantification of total cell and neuron numbers in the brain. Journal of Neuroscience, 25(10), 2518–2521. DOI: 10.1523/JNEUROSCI.4526-04.2005
Herculano-Houzel, S. (2009). The human brain in numbers: a linearly scaled-up primate brain. Frontiers in Human Neuroscience, 3, 31. DOI: 10.3389/neuro.09.031.2009
Herculano-Houzel, S. (2012). The remarkable, yet not extraordinary, human brain as a scaled-up primate brain and its associated cost. PNAS, 109(Suppl 1), 10661–10668. DOI: 10.1073/pnas.1201895109
Von Bartheld, C. S., Bahney, J. & Herculano-Houzel, S. (2016). The search for true numbers of neurons and glial cells in the human brain: A review of 150 years of cell counting. Journal of Comparative Neurology, 524(18), 3865–3895. DOI: 10.1002/cne.24040
Pakkenberg, B. & Gundersen, H. J. G. (1997). Neocortical neuron number in humans: Effect of sex and age. Journal of Comparative Neurology, 384(2), 312–320. PMID: 9215725
Goriely, A. (2025). Eighty-six billion and counting: do we know the number of neurons in the human brain? Brain, 148(3), 689. DOI: 10.1093/brain/awae390
Bahney, J. & von Bartheld, C. S. (2014). Validation of the isotropic fractionator: comparison with unbiased stereology and DNA extraction for quantification of glial cells. Journal of Neuroscience Methods, 222, 165–174. PMID: 24239779
Siletti, K., Hodge, R. et al. (2023). Transcriptomic diversity of cell types across the adult human brain. Science, 382(6667). DOI: 10.1126/science.add7046
Shapson-Coe, A. et al. (2024). A petavoxel fragment of human cerebral cortex reconstructed at nanoscale resolution. Science, 384(6696). DOI: 10.1126/science.adk4858
Goriounova, N. A. et al. (2018). Large and fast human pyramidal neurons associate with intelligence. eLife, 7, e41714. DOI: 10.7554/eLife.41714
Pakkenberg, B. et al. (2020). Is there a correlation between the number of brain cells and IQ? Cerebral Cortex, 31(1), 650–657. DOI: 10.1093/cercor/bhaa249
Herculano-Houzel, S. et al. (2014). The elephant brain in numbers. Frontiers in Neuroanatomy, 8, 46. DOI: 10.3389/fnana.2014.00046