Viking-era researchers concluded that impact craters of diameter D<50 m were absent on Mars, and thus impact gardening was considered negligible in establishing decameter-scale surface properties. This paper documents martian crater populations down to diameter D∼11 m and probably less on Mars, requiring a certain degree of impact gardening. Applying lunar data, we calculate cumulative gardening depth as a function of total cratering. Stratigraphic units exposed since Noachian times would have experienced tens to hundreds of meters of gardening. Early Amazonian/late Hesperian sites, such as the first three landing sites, experienced cumulative gardening on the order of 3–14 m, a conclusion that may conflict with some landing site interpretations. Martian surfaces with less than a percent or so of lunar mare crater densities have negligible impact gardening because of a probable cutoff of hypervelocity impact cratering below D∼1 m, due to Mars' atmosphere. Unlike lunar regolith, martian regolith has been affected, and fines removed, by many processes. Deflation may have been a factor in leaving widespread boulder fields and associated dune fields, observed by the first three landers. Ancient regolith provided a porous medium for water storage, subsurface transport, and massive permafrost formation. Older regolith was probably cemented by evaporites and permafrost, may contain interbedded sediments and lavas, and may have been brecciated by later impacts. Growing evidence suggests recent water mobility, and the existence of duricrust at Viking and Pathfinder sites demonstrates the cementing process. These results affect lander/rover searches for intact ancient deposits. The upper tens of meters of exposed Noachian units cannot survive today in a pristine state. Intact Noachian deposits might best be found in cliffside strata, or in recently exhumed regions. The hematite-rich areas found in Terra Meridiani by the Mars Global Surveyor are probably examples of the latter.