This study aims at the comparison of different site-effect estimation techniques. We used aftershock records of the 14 August 2003 (Mw 6.2) earthquake that occurred in the western part of Greece next to Lefkas Island. A data set of 15 events recorded at five stations was selected for the period of 20 August 2003–30 September 2003. The moment magnitudes range between 3.3 and 4.2, and the hypocentral distances vary from 5 to 34 km. The site effects are determined using the standard spectral ratio (SSR) method, the horizontal-to-vertical spectral ratio for earthquake data method (the so-called receiver function [RF] method), and an inversion method for simultaneous source, path, and site-effects estimation. These methods are applied to both P and S waves, using exactly the same time windows in order to compute the spectra. Finally, using geotechnical information available for the city of Lefkas, the experimental site transfer functions at each site are compared to theoretical one-dimensional 1D transfer functions computed using Kennett’s reflectivity coefficient method. We show that the SSR method and the inversion method give similar site transfer functions. However, the experimental SSR fits even better with the spectral ratios of the inversion results, which indicates that our reference site is affected by a small site effect. Moreover, the RF method results fit well with the horizontal-to-vertical ratios coming from the inversion results, indicating in this case a nonnegligible amplification on the vertical component. This is the reason why the RF gives only the frequency of amplification but not the amplitude. These conclusions hold for both P and S waves. Finally, the site effects computed using the whole record or a window including only S waves are similar, while the P waves results, although showing some similar features with those for S waves, give different transfer functions for the sites investigated.