Crater counts estimate how long a surface has been exposed to impacts and has retained their marks. Under comparable conditions, a surface with more impact craters per unit area is generally older than one with fewer. Turning that comparison into an age in years requires a model of impact production, calibration and a careful account of resurfacing.
The result concerns a surface’s history, not automatically the age of its planet or the rock deep beneath it. Lava can bury an old landscape and start a younger record on top. Erosion, burial and later impacts can also change which craters survive. The Planetary Science Institute’s crater chronology explanation calls attention to this distinction through the concept of crater retention age.
Count an area, not just a collection of circles
Begin by defining a geologic unit: a stretch of terrain with a history that can reasonably be analyzed together. Mixing an ancient cratered highland with a neighboring young lava plain can produce a number that describes neither one well.
Next, define which crater sizes can be identified consistently in the available images. A map that resolves small features and one that misses them cannot be compared as if both inventories were complete. A 2025 study cataloged by the USGS specifically examines how resolution and viewing and illumination conditions affect crater detection; its abstract supports treating detectability as part of the measurement, rather than an afterthought.
For an example of a stated size threshold, NASA’s 2010 Lunar Reconnaissance Orbiter crater census described more than 5,000 lunar craters at least 20 kilometers across. That was a particular survey of large craters, not a claim to have counted every lunar crater or a current total for the Moon.
A larger raw count can belong to the less cratered surface
Consider two hypothetical regions mapped with equally suitable data. In both, count only impact craters at least one kilometer across.
Region A contains 20 such craters in 1,000 square kilometers. Its crater density is 20 / 1,000 = 0.020 per square kilometer. Region B contains 40 in 4,000 square kilometers: 40 / 4,000 = 0.010 per square kilometer.
Region B has twice as many craters in total, but Region A has twice the number per unit area. Comparing the raw counts alone would reverse the useful comparison.
Even the normalized result does not prove that A is exactly twice as old. That would require extra assumptions about the impact rate, crater preservation and the applicable age range. These invented counts demonstrate normalization; they are not observations or dates for real terrain.
Actual studies commonly examine a size–frequency distribution, recording how crater abundance changes across diameter ranges. A count that agrees with one simple age interpretation at large diameters but not at small ones can signal a more complicated surface history.
Where an age in years comes from
Relative ordering needs fewer assumptions than an absolute date. Superposition can show that one feature lies on top of another. NASA’s Copernicus region explanation combines those geologic relationships with crater abundance to discuss older and younger surfaces.
To attach years, researchers need calibration. Returned lunar samples allow laboratory age measurements to be connected with mapped terrain. NASA’s Apollo 17 geology account explains both the importance of samples and a central difficulty: an impact can move rocks, so a sample collected in one place may have originated elsewhere.
The dated event in a rock must therefore be related correctly to the surface being counted. A rock’s age does not automatically timestamp every surrounding landform. The association between sample, geologic unit and the event that established its visible surface is part of the inference.
Applying a chronology to Mars or another body also requires physical modeling. The impactor population, encounter speeds, gravity and target properties affect crater production. The PSI discussion explicitly examines why transferring a lunar production relationship to Mars is more complicated than multiplying all crater sizes by one universal factor. Its historical model iterations illustrate the method, rather than supplying a single final age scale for every modern study.
Three ways the record can mislead
Resurfacing can erase part of the clock. A lava flow may bury small craters while some large structures remain recognizable. An inventory can then contain traces of more than one event. “The surface is old” needs a definition of which surface and which retained crater population are being dated.
One impact can create many secondary craters. Ejected material can fall back and excavate additional pits. Those secondaries are not independent arrivals from the background impactor population. JPL’s HiRISE example of secondary craters shows why clustering and spatial patterns matter when identifying a population.
An intensely cratered surface can approach saturation. New impacts destroy or overlap older craters. Beyond that point, surviving crater abundance does not keep increasing as a simple linear clock. The PSI chronology discussion describes this limitation; dense cratering is not permission to extrapolate an unlimited age from a raw count.
These issues explain why a serious crater age comes with a defined terrain boundary, diameter range, model and uncertainty. They also make the method more useful: the mismatches can reveal resurfacing or mixed histories rather than merely spoil a calculation.
Our guide to planets and moons places cratered terrain alongside volcanism and other geologic processes. Crater counting adds a specific tool to that comparison: a calibrated, incomplete record of exposure and preservation, not a birth certificate for an entire world.
AI-generated illustration for Galileo Whispers of a generic cratered airless surface adjoining a smoother plain. Terrain reference: NASA/GSFC/Arizona State University. This is not Copernicus, a survey image or a measured age map.
