Very low frequency (VLF) data recorded in the midwestern United States are compared with sprite observations on three different dates of high sprite activity between 1995 and 2000. Sprites are frequently seen to be correlated with "early/fast" events: perturbations of the amplitude and/or phase of a VLF transmitter signal propagating through the storm region. Unlike recent observations of Haldoupis et al. (2004), where a one-to-one relationship between forward scatter events and sprites was reported, only ~48% of sprites in our cases are accompanied by VLF perturbations, or, viewed conversely, ~61% of VLF perturbations are accompanied by sprites. The difference between the two data sets might be due to the relative location of the causative lightning along the VLF signal path (i.e., near transmitter or near receiver) in that mode coupling of the propagating VLF signal causes some perturbations to be undetectable at long distances from the disturbed region. In addition, there are quite a few cases of early/fast perturbations not accompanied by sprites, although not nearly the preponderance previously expected. In rare cases, observed VLF signal perturbations can only be explained by VLF backscatter from the sprite body because of the geometrical constraints that preclude other mechanisms, including forward scattering from elves or halos. However, such backscatter events are extremely rare and are found to occur only for the largest sprites in terms of their horizontal extent.