Narrow, fan-shaped dark streaks on steep Martian slopes were originally observed in Viking Orbiter images, but a definitive explanation was not possible because of resolution limitations. Pictures acquired by the Mars Orbiter Camera (MOC) aboard the Mars Global Surveyor (MGS) spacecraft show innumerable examples of dark slope streaks distributed widely, but not uniformly, across the brighter equatorial regions, as well as individual details of these features that were not visible in Viking Orbiter data. Dark slope streaks (as well as much rarer bright slope streaks) represent one of the most widespread and easily recognized styles of mass movement currently affecting the Martian surface. New dark streaks have formed since Viking and even during the MGS mission, confirming earlier suppositions that higher contrast dark streaks are younger, and fade (brighten) with time. The darkest slope streaks represent ∼10% contrast with surrounding slope materials. No small outcrops supplying dark material (or bright material, for bright streaks) have been found at streak apexes. Digitate downslope ends indicate slope streak formation involves a ground-hugging flow subject to deflection by minor topographic obstacles. The model we favor explains most dark slope streaks as scars from dust avalanches following oversteepening of air fall deposits. This process is analogous to terrestrial avalanches of oversteepened dry, loose snow which produce shallow avalanche scars with similar morphologies. Low angles of internal friction typically 10–30¡ for terrestrial loess and clay materials suggest that mass movement of (low-cohesion) Martian dusty air fall is possible on a wide range of gradients. Martian gravity, presumed low density of the air fall deposits, and thin (unresolved by MOC) failed layer depths imply extremely low cohesive strength at time of failure, consistent with expectations for an air fall deposit of dust particles. As speed increases during a dust avalanche, a growing fraction of the avalanching dust particles acquires sufficient kinetic energy to be lost to the atmosphere in suspension, limiting the momentum of the descending avalanche front. The equilibrium speed, where rate of mass lost to the atmosphere is balanced by mass continually entrained as the avalanche front descends, decreases with decreasing gradient. This mechanism explains observations from MOC images indicating slope streaks formed with little reserve kinetic energy for run-outs on to valley floors and explains why large distal deposits of displaced material are not found at downslope streak ends. The mass movement process of dark (and bright) slope streak formation through dust avalanches involves renewable sources of dust only, leaving underlying slope materials unaffected. Areas where dark and bright slope streaks currently form and fade in cycles are closely correlated with low thermal inertia and probably represent regions where dust currently is accumulating, not just residing.