Getting Accurate Stereotaxic Coordinates Right
Stereotaxic surgery on mice is straightforward until it isn't, and most of the time it isn't because of small coordinate errors that compound across multiple animals. The system works by referencing a point on the skull—bregma—and measuring relative distances along three axes: anterior-posterior (AP), medial-lateral (ML), and dorsal-ventral (DV). That sounds simple enough on paper. In practice, there are several things that will push your injection site or recording electrode millimeters off target, and millimeters in the mouse brain is the difference between hitting the substantia nigra and hitting white matter above it. Most labs pull their reference data from the Franklin and Paxinos mouse brain atlas, which is available as a free PDF download from the University of Toronto's website. There's also the Allen Brain Atlas with interactive 3D viewers and the Dorr et al. 2008 atlas for C57BL/6J specifically. These are the three I use interchangeably depending on the experiment. The Franklin and Paxinos book is the standard reference most protocols cite, but if you're working with viral vectors or optogenetics targeting specific subnuclei, the Allen atlas gives you better resolution for cortical layer work. I've also used the Deacon atlas for developmental studies since it covers a wider age range. Download links for these are scattered across university lab sites and academic repositories. The Franklin-Paxinos atlas PDF is roughly 40MB and contains coronal plates every 100 microns for adult mice. That level of detail matters when you're targeting something like the ventral tegmental area, which spans roughly AP -3.0 to -3.8 from bregma depending on the subregion.
The Coordinate System Itself
The origin point is bregma, the junction where the coronal and sagittal sutures meet. AP is positive rostrally and negative caudally. ML is positive to the right and negative to the left in standard right-handed systems, though some labs flip this convention so double-check whatever atlas you're using. DV is positive from the brain surface downward toward the ventricles. You measure from the skull surface, not from the top of the head, which means any scalp tension or swelling during surgery shifts your effective zero point. Another thing most people miss is that the interaural line isn't perfectly flat on a living mouse. The skull curves, and if you don't level the ears properly, your DV measurements drift. I had a postdoc once who spent three weeks troubleshooting why hCZ viral injections were landing in the cerebellum instead of the hippocampus. The problem was that her stereotaxic frame had a worn ear bar on the left side, and she'd been injecting at ML +1.2 when the actual target was closer to ML +0.8. She corrected it by checking the skull surface directly with a microdrive instead of relying on the frame's default leveling. That single adjustment brought her success rate from about 40 percent to over 85 percent in subsequent experiments.
Practical Considerations When Using the Coordinates
Body weight matters more than people usually account for. A 20-gram mouse and a 30-gram mouse have different skull sizes and thus different absolute coordinate values even when targeting the same brain region. Most atlases give coordinates for a 25-gram mouse, so if your animals are outside that range, you'll need to scale them. There's no clean linear scaling formula, but a rough adjustment of about 2 to 3 percent per gram deviation is what I typically use as a starting point. Anesthesia depth affects your DV readings. Deep isoflurane causes cerebral vasodilation and slight brain swelling, which pushes the target deeper relative to the skull surface. If you're doing electrophysiology recordings, you'll want lighter anesthesia to minimize this effect. For viral injections, slightly deeper anesthesia is actually preferable because it reduces movement artifacts and bleeding. This trade-off means the same coordinate list serves two different purposes depending on your end goal. One specific edge case that catches people off guard involves the lambda reference point. Lambda sits at the junction of the lambdoid and sagittal sutures, roughly 5.5 mm caudal to bregma in adult C57BL/6 mice. Some surgeons prefer to level using both bregma and lambda as a check. If those two points aren't on the same horizontal plane, the skull isn't level and your entire coordinate set is compromised. I check this constantly because even a half-degree tilt across the skull translates to about 0.1 mm of error at typical injection sites, which is negligible for gross targeting but unacceptable when you're trying to hit a nucleus that's only 0.5 mm tall.
Get the Full Details
A Note on Atlas Limitations
No single atlas covers every situation. The Franklin and Paxinos atlas is based on a handful of C57BL/6 mice and doesn't account for strain differences in brain morphology. BALB/c mice have proportionally larger lateral ventricles, which shifts midline-adjacent targets. If you're using a transgenic line with enlarged or shrunken ventricles due to the construct, standard coordinates won't reliably land you in the right place. I've seen papers where researchers used wild-type atlas coordinates on a Cre-driver line and got completely inconsistent expression patterns as a result. Also worth noting: stereotaxic coordinates assume a rigid skull-brain relationship. In neonatal mice, the fontanelles are still open, and the skull moves independently of the brain. Adult mice are fine, but anything under postnatal day 10 requires a fundamentally different approach. Some labs use magnetic resonance imaging-guided targeting for neonatal work instead, which is more accurate but significantly more expensive and time-consuming.
What I Actually Do Before Starting a Surgery
I pull up the relevant coronal sections from whichever atlas fits my strain and target structure, note the AP, ML, and DV values, and then add a 0.2 to 0.3 mm DV safety margin depending on whether I'm injecting or recording. For injections, I go slightly deeper than the coordinate suggests because tissue compression from the needle track can shift the deposited volume upward by a fraction of a millimeter. For electrophysiology, I stop 0.2 mm short of the target coordinate to account for the meninges and pia, which add roughly that much distance between the skull surface and the actual gray matter. I also run a quick sanity check against a second atlas. If Paxinos says AP -2.8 and the Allen atlas says AP -2.5 for the same structure, I average them rather than trusting either one blindly. These discrepancies exist because different labs used different histological methods and different numbers of animals. Taking the midpoint usually lands you in the right neighborhood without committing to an outlier value.
