Surface-wave methods are quite popular for site characterization in geotechnical earthquake engineering. Among these techniques, a particular role is taken by passive methods for their ability to yield information on the low-frequency range and consequently on large depths. One such passive method, the refraction microtremors (ReMi) technique, has been proposed as a simple alternative to 2D-array techniques to estimate surface-wave dispersion by using linear arrays of geophones. The technique owes its name to the use of widely available instruments also adopted for seismic refraction. The basic hypotheses underlying ReMi are that noise is distributed isotropically in azimuth or is aligned exactly with the array. These conditions often are not met, and in most cases they are not verified because such analysis requires an accurate approach to data processing that is rarely applied. We have developed an algorithm that verifies ReMi's basic hypotheses by analyzing experimental data. In addition, we have proposed an algorithm to identify the lowest apparent velocity on the ReMi spectra, thus avoiding interpretation problems.