Precipitation data from all Defense Meteorological Satellite Program (DMSP) satellites over a 9-year interval were surveyed for electron acceleration events. Selected events evinced a sharp drop above the spectral peak; therefore only instances where the accelerating potential is 3--4 or more times larger than the source thermal energy were studied. In all, more than 1.5¿108 individual electron spectra were examined for signs of such substantial field-aligned acceleration. Several controversies can be resolved by this survey; for example, the ''midday gap'' is real, with noon easily the least likely local time for observing electron acceleration. The debated ''14 MLT (magnetic local time) hot spot'' is also real, although it is centered about 15 MLT and is distinct from the rest of the oval only for northward interplanetary magnetic field (IMF) conditions. A weaker ''warm'' spot exists in the 6--9 MLT region. Potential drops of ~3 keV and above are statistically the province of the nightside, as are the most intense energy fluxes (>1 erg cm-2 s-1). The latitudinal scale size of acceleration events fits surprisingly well to an exponential distribution with a characteristic width (⟨w⟩=28--35 km) that varies only slightly with MLT and IMF conditions. If ⟨w⟩ represents the ionospheric footprint of the driving magnetospheric processes, the best candidate explanation is velocity shears in the main plasma sheet. However, the fact that an exponential distribution of latitudinal widths exists under a variety of conditions suggests an alternative. If there exist large-scale regions embedded within the auroral oval wherein each spectrum has a random probability P of experiencing electron acceleration independent of neighboring spectra, then the occurrence distributions are explained, and ⟨w⟩ actually indicates the value of P (namely, ~0.85). ¿ American Geophysical Union 1996