A survey of environmental seismic noise (microtremors) was carried out within the Catania municipal area in May 1999. The aim of the data acquisition was to improve our prediction of the seismic ground motion locally, using Nakamura\'s approach. The seismic noise was recorded at 39 different sites. These sites were chosen according to the following criteria: 1) significance in terms of geological, geotechnical, or local characteristics; 2) presence of a lithological transition (e.g.: the transition from lava to sediments); 3) presence of other geotechnical or geophysical measurements; 4) alignment with some of the transects along which the ground motion had been simulated numerically. The data were acquired independently by two teams, who used different instrumentation. Measurements overlapped at seven locations. We have found that more than one half of the sites exhibit weak or no amplification. These sites are located either on lava (e.g., sites in downtown Catania and in the northern part of the municipal area) or on well-consolidated sedimentary soils (Western districts of the city). The only sites which bear evidence of some amplification are located either on soil fill lying over lava, or on the fine alluvial deposits of the Catania Plain. In the first case the H/V amplitude is usually moderate and the fundamental frequency is between 5 and 12 Hz, while the sites of the Plain have a much lower natural frequency (about 1.3 Hz) and larger H/V amplitude. Some sites displayed anomalous, though relatively small in most cases, H/V peaks, in the low frequency band 0.2-1 Hz. Our analysis, performed at some sites, comparing the measured H/V spectral ratios (HVSR) to those simulated numerically for an earthquake, supports the interpretation raised recently by other authors of an effect of the deep structure. Finally, the numerical simulations also predict an appreciable lateral variability of the HVSR, where the underlying structure is not a \'plane layer model\', rather usual in Catania. This implies that seismic noise measurements made at single, isolated points should be used with care in the interpretation.