Coalescing compact binaries have been pointed out as the most promising source of gravitational waves for kilometer-size interferometers such as LIGO. Gravitational wave signals are extracted from the noise in the detectors by matched filtering. This technique performs really well if ana prioritheoretical knowledge of the signal is available. The information known about the possible sources is used to construct a model of the expected waveforms (templates). A common assumption made when constructing templates for coalescing compact binaries is that the companions move in a quasi-circular orbit. Some scenarios, however, predict the existence of eccentric binaries. We investigate the loss in signal-to-noise ratio due to non-optimal filtering of eccentric signals and calculate the fitting factor associated with the matched filtering of eccentric signals by circular templates. ©1999 American Institute of Physics.