Existing quartz cement models assume that the rate of growth per unit surface area is independent of grain size. Application of one such model to four geologically diverse data sets reveals a systematic error with grain size such that values in finer grained sandstones are overpredicted. Our laboratory synthesis of quartz overgrowths indicates that this grain-size effect results from the more rapid development of euhedral crystal forms on smaller grains. Experiments show that the rate of growth along the quartz c axis drops by a factor of about 20 after euhedral faces develop. Our numerical simulations of quartz growth in two dimensions indicate that this euhedral effect should be significant in sandstones despite the complexity that arises from the interaction of multiple growing crystals and small pore sizes. Simulations also suggest that this phenomenon is responsible for the common observation that quartz overgrowths are less extensively developed on chert and polycrystalline grains compared to monocrystalline grains. This euhedral effect may also explain the common observation that quartz growth rates are significantly faster on fracture surfaces compared to detrital grain surfaces. Most sand grains have well-developed dust rims that reflect minor adhesions of nonquartz materials or damage from surface abrasions or impacts. Our numerical and laboratory experiments indicate that such small-scale discontinuities dramatically reduce initial rates of quartz growth because they break overgrowths into separate smaller crystal domains that are bounded by euhedral faces. The paucity of nucleation discontinuities on fracture surfaces should lead to substantially faster rates of growth compared to grain surfaces.