30,000 neurons of 70 million

What you are looking at

The outline is the Allen Institute's reference mouse brain, the Common Coordinate Framework, an average of many brains that every dataset here was registered to by the people who made it. That shared frame is the whole trick: neurons traced in different labs, in different years, with different microscopes, land in the same brain and can be drawn together.

Each neuron was filled with a fluorescent protein, the whole brain was imaged at sub-micron resolution, and a person or an algorithm followed every branch of the axon through the stack, sometimes for centimetres. Six collections are drawn: the Janelia MouseLight project, the SEU-ALLEN collaboration between Southeast University and the Allen Institute, three projectome atlases from the Institute of Neuroscience in Shanghai, of the prefrontal cortex, the hippocampus and the whole cortex, and the 132 locus coeruleus noradrenaline neurons of Su et al. 2026 from the Allen Institute for Neural Dynamics, the longest mouse neurons ever traced. The counts on the card are what loaded, not a number typed in.

The locus coeruleus, one way in. A speck of about 1,500 cells in the pons is the brain's main source of noradrenaline, and each of its cells reaches a wide but particular set of regions. Su et al. 2026 traced 132 of them whole, and found that a cell body's depth in the nucleus decides where its axon goes: the dorsal cells to the cortex, the ventral ones to the brainstem and spinal cord. See all 132 at once, coloured by depth, or pick the dataset below and light them one at a time. Axon lengths on their cards are measured on the released trees; the paper reports a mean of 34 cm and a longest of 70 cm.

A sample, not the population. A mouse brain has about 70 million neurons and this page draws about thirty-two thousand. It shows where these neurons go. It says nothing about how many neurons go there, and the divisions with the most cells here are the ones these projects chose to label, not the ones with the most neurons.

Signal speed

The light that runs the axon is a stand-in for an action potential. It travels along the tree at 4 millimetres per second of cable, measured from the cell body. A real spike in a thin unmyelinated axon covers about a metre per second, and a myelinated one several times that, so what you see is slowed roughly 250 times. The card gives each neuron's real crossing time at one metre per second.

What this is not

An axon is not a connection Each line is where one axon goes. Where it makes synapses, and onto which cells, is not in this data. The bar on the card counts axon tips by division, which is the honest proxy available.

Not a whole-brain connectome Nobody has one for a mouse yet. The most complete mouse connectome at synapse level is MICrONS: every synapse in one cubic millimetre of visual cortex, about 75,000 neurons and half a billion synapses. Turn on the cubic millimetre to see that block at its true size, in the visual cortex where it came from.

The human page on this site draws the opposite thing. On the white matter each cable is a bundle of a million axons averaged over a thousand people, because no one can trace a single axon across a human brain. Here each line is one axon in one mouse.

Every cell

The every cell layer is a different kind of data: the Allen Brain Cell Atlas imaged one mouse brain in serial sections and recorded, for every cell, which of about a thousand genes it was expressing. That gives a position and a cell type for millions of cells in one animal, also in the reference frame. The page draws one cell in twelve, coloured by the atlas's own 34 cell classes and its own colours. Because the brain was cut into sections a fifth of a millimetre apart, the cloud shows a faint striping at the section spacing. That is the measurement, not an artefact of drawing.

Where the data comes from

How many neurons a mouse has. Herculano-Houzel S., Mota B. and Lent R., Cellular scaling rules for rodent brains, PNAS 103, 12138 (2006): about 71 million neurons in the whole mouse brain, counted by isotropic fractionation. The title rounds both numbers.
Reference brain. Allen Mouse Brain Common Coordinate Framework, version 3, 2017 annotation and structure meshes. Wang Q. et al., Cell 181, 936 (2020). Allen Institute terms of use.
Locus coeruleus. Su Z., Kosillo P., Jung K. et al., Topographic structure and function of locus coeruleus noradrenaline neurons, Nature (2026), 10.1038/s41586-026-11026-0, CC BY 4.0. 132 whole-brain reconstructions from 4 Dbh-Cre mice, imaged by ExA-SPIM and traced in HortaCloud at the Allen Institute for Neural Dynamics, read from the public aind-open-data bucket in CCF v3 coordinates, with the authors' locus coeruleus surface (the densest 67% of Dbh+ cell bodies pooled from 8 brains). Each cell's class here is its primary projection target and its depth is its cell body's rank along the dorsal to ventral axis, both computed from the released trees; the spinal cord is outside the reference brain, so a spinal projection shows as axon leaving through its caudal face.
MouseLight. Winnubst J. et al., Reconstruction of 1,000 projection neurons reveals new cell types and organization of long-range connectivity in the mouse brain, Cell 179, 268 (2019). Data from the MouseLight NeuronBrowser, Janelia Research Campus, CC BY-NC 4.0, CCF v3.0 coordinates.
SEU-ALLEN. Peng H. et al., Morphological diversity of single neurons in molecularly defined cell types, Nature 598, 174 (2021). 1,741 registered reconstructions from the Brain Image Library, CC BY-SA 4.0.
ION prefrontal cortex. Gao L. et al., Single-neuron projectome of mouse prefrontal cortex, Nature Neuroscience 25, 515 (2022). 6,357 reconstructions from the Digital Brain projectome atlas, Institute of Neuroscience, Chinese Academy of Sciences, Shanghai. Dataset DOI 10.12412/BSDC.1690164952.20001.
ION hippocampus. Qiu S. et al., Whole-brain spatial organization of hippocampal single-neuron projectomes, Science 383, eadj9198 (2024). 10,100 reconstructions from the Digital Brain hippocampus atlas, same institute. Dataset DOI 10.12412/BSDC.1667278800.20001. The Shanghai reconstructions carry the axon and the cell body; dendrites were traced separately and are not drawn here.
ION whole cortex. Gao L. et al., Integrative analysis of single-neuron projectomes links connectome, transcriptome, and function in the mouse cortex, Neuron 114, 86 (2026). 18,621 reconstructions across the whole cortex from the Digital Brain whole-cortex atlas, same institute, with the portal's projectome classes (IT, PT and CT groups). 6,357 of them are the prefrontal reconstructions above under the same ids, so 12,264 are new here; the prefrontal set keeps its own class scheme.
Every cell. Yao Z. et al., A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain, Nature 624, 317 (2023). Allen Brain Cell Atlas, MERFISH dataset C57BL6J-638850, CCF coordinates release 2023-12-15, CC BY-NC 4.0.
The cubic millimetre. MICrONS Consortium, Functional connectomics spanning multiple areas of mouse visual cortex, Nature 640, 435 (2025). Shown here only as a box at its size, placed at the centroid of primary visual cortex in the reference brain, which is where that volume was taken to within what this page can show.

Where an axon's tips land is computed by this page from the 2017 annotation volume at 25 µm, tip by tip, and reported at the level of the twelve major divisions. Cell body regions are looked up the same way and were checked against the regions Janelia lists for every MouseLight neuron. The build script, tools/build-mouse.py, and its checks ship with the site's source.