It is well known that surface geology may significantly affect strong ground motion and hence it should be incorporated into seismic hazard estimation at a specific site. Numerical and empirical methods are used to determine the site effect. On areas with moderate seismicity the empirical methods are based on microtremor measurements. Solely the single station methods, for example the H/V spectral ratio technique, could be used in the absence of a nearby reference station situated on rock site. The one year long noise measurements of 8 stations belonging to the Hungarian Microseismic Monitoring Network have made it possible the testing of this method. The stations are situated in the middle part of Hungary. All seismometers are triaxial LE-3D high sensitivity 1 Hz geophones. Three stations are situated on hard rocks while five ones on loose site above a deep sedimentary basin with more than 1000m thick sediments. The large amount of noise data have made it possible to examine the stability of the method. The examinations have shown that the computed spectral ratio curves are very stable in time. While all stations on rock sites show nearly at H/V spectral ratio, all stations on soft sites show a peak at very low, 0.1- 0.2Hz frequencies. Analytical computations have been carried out to verify this low frequency behaviour and have shown that these peaks are due to the fundamental frequency of the overall sediment thickness. H/V ratios have also been calculated from seismograms of weak local earthquakes, so they could be compared with ratios obtained from noise measurements.