Intraplate earthquakes are characterized by spatially isolated regions of persistent, diffuse earthquake activity with no general correlation with mapped surface geology. Various models have been proposed for their origin, but there has been no widely accepted explanation. Here I show a spatial correlation between surface streamflow attributes (mean annual streamflow, standard deviation of streamflow) and the geographic distribution of epicenters in Virginia, West Virginia, and New Madrid, Missouri, supporting the hypothesis that rainfall plays a key role in the generation of intraplate earthquakes. Fluctuations in the elevation of the earth-atmosphere interface (water table, reservoir-surface, river stage) are associated with pore-fluid pressure transients that are transmitted to hypocentral depths where they trigger earthquakes. Such a meteorologically and topographically driven model has been referred to as “hydroseismicity.” The new results link flow in major rivers to the distribution of epicenters and reveal a clear correlation between the supply of water at the Earth's surface and the areal distribution of earthquake epicenters. In addition, the timing of three large intraplate earthquakes in Virginia and West Virginia with respect to hurricanes Camille and Agnes, together with the known or inferred values of crustal hydraulic diffusivity, suggests a causal relationship between the meteorologic impulses and the earthquakes.