Many shallow craters near the Spirit Mars Exploration Rover landing site contain asymmetric deposits of windblown sediments which could indicate the predominant local wind direction at the time of deposition or redistribution. Wind tunnel simulations and field studies of terrestrial craters were used to determine trends in deposition as a function of crater morphometry and wind direction. Terrestrial analog field work at the Amboy lava field, Mojave Desert, California, included real-time wind measurements and assessments of active sediment deposition in four small (<100 m) craters. Preliminary results indicate that reverse flow or stagnant wind and deposition on the upwind side of the crater floor occurs in craters with depth-to-diameter (d/D) ratios ≥0.05. Measurements taken within a crater of d/D of ~0.02 do not indicate reverse flow. Therefore, reverse flow is expected to cease within a d/D range of 0.02 to 0.05, resulting in wind movement directly over the crater floor in the downwind direction with no asymmetric sediment deposition. Wind tunnel simulations using six crater models, including a scaled model of a crater from the Amboy lava field, were completed to assess the wind flow in and around craters as a function of crater morphometry (depth, diameter). Reverse flow occurred in craters with d/D ratios ≥0.033, resulting in sediment deposition in the upwind portion of the crater floor. Visual observations of a crater with a d/D of ~0.020 did not indicate reverse flow, similar to the results of field studies; therefore, reverse flow appears to cease within a d/D range of 0.020 to 0.033. Craters with asymmetric aeolian deposits near the Mars Spirit landing site have d/D ratios of 0.034 to 0.076, suggesting that reverse flow occurs in these craters. Thus, the position of windblown sediments in the northwest parts of the crater floors would indicate prevailing winds from the northwest to the southeast, consistent with late afternoon winds as predicted by the Mars Regional Atmospheric Modeling System (MRAMS) circulation model.