We investigate the feasibility of using microtremor measurements in short and long period ranges for the prediction of frequency-dependent amplification effects due to local site geology in the Granada basin and Granada city (Southern Spain). Following a seismic reflection profile crossing the Granada basin, microtremor measurements were recorded at 12 different sites. Peaks centered at 0.7 s (1.5 Hz) and 3-4 s are observed in the microtremor spectra. The 0.7 s peak is interpreted in terms of microtremors and the second one in terms of ocean-generated microseisms. The former is present at all sites on Pliocene and Quaternary sedimens while the 3-4 s peak is present at all sites regardless of the surface geology, although with differences in the spectral amplitude. In order to investigate the possible influence of geologic parameters on long period motion, the maximum spectral amplitude value in the range 1 to 5 s was estimated at all sites in Granada basin. A good linear relation between the maximum spectral amplitude value and surface geology and depth to underground basement was found. However, for short period range (T<1 s), only the surface geology has any relation with the maximum spectral amplitude value. For Granada city, differences in the microtremor spectra were found to depend on the local geology. In the downtown area, situated on alluvial clay and sand from the Quaternary age, differences of up to 10 times with respect to the hill zones (located on compact conglomerates of Pliocene age) are observed. Results from coda-Q analysis from local earthquakes in the Granada basin show overestimations of Q_c, for lapse times of 70 s, of up to 4 times at stations situated on sediment when compared to those obtained on bedrock for the 1.5 Hz frequency and to a lesser extent for 3 Hz. This overestimation is explained in terms of seismic-wave amplification in the Granada basin.