A 10-km diameter crater named Zunil in the Cerberus Plains of Mars created ∼107 secondary craters 10 to 200 m in diameter. Many of these secondary craters are concentrated in radial streaks that extend up to 1600 km from the primary crater, identical to lunar rays. Most of the larger Zunil secondaries are distinctive in both visible and thermal infrared imaging. MOC images of the secondary craters show sharp rims and bright ejecta and rays, but the craters are shallow and often noncircular, as expected for relatively low-velocity impacts. About 80% of the impact craters superimposed over the youngest surfaces in the Cerberus Plains, such as Athabasca Valles, have the distinctive characteristics of Zunil secondaries. We have not identified any other large (⩾10 km diameter) impact crater on Mars with such distinctive rays of young secondary craters, so the age of the crater may be less than a few Ma. Zunil formed in the apparently youngest (least cratered) large-scale lava plains on Mars, and may be an excellent example of how spallation of a competent surface layer can produce high-velocity ejecta (Melosh, 1984, Impact ejection, spallation, and the origin of meteorites, Icarus 59, 234–260). It could be the source crater for some of the basaltic shergottites, consistent with their crystallization and ejection ages, composition, and the fact that Zunil produced abundant high-velocity ejecta fragments. A 3D hydrodynamic simulation of the impact event produced 10¹⁰ rock fragments ⩾10 cm diameter, leading to up to 10⁹ secondary craters ⩾10 m diameter. Nearly all of the simulated secondary craters larger than 50 m are within 800 km of the impact site but the more abundant smaller (10–50 m) craters extend out to 3500 km. If Zunil is representative of large impact events on Mars, then secondaries should be more abundant than primaries at diameters a factor of ∼1000 smaller than that of the largest primary crater that contributed secondaries. As a result, most small craters on Mars could be secondaries. Depth/diameter ratios of 1300 small craters (10–500 m diameter) in Isidis Planitia and Gusev crater have a mean value of 0.08; the freshest of these craters give a ratio of 0.11, identical to that of fresh secondary craters on the Moon (Pike and Wilhelms, 1978, Secondary-impact craters on the Moon: topographic form and geologic process, Lunar Planet. Sci. IX, 907–909) and significantly less than the value of ∼0.2 or more expected for fresh primary craters of this size range. Several observations suggest that the production functions of Hartmann and Neukum (2001, Cratering chronology and the evolution of Mars, Space Sci. Rev. 96, 165–194) predict too many primary craters smaller than a few hundred meters in diameter. Fewer small, high-velocity impacts may explain why there appears to be little impact regolith over Amazonian terrains. Martian terrains dated by small craters could be older than reported in recent publications.