The main goal of this thesis is to test the brittle-ductile (BD) interaction hypothesis of the lithosphere. The model uses the correlation of two parameters, the seismicity characterizing the brittle part and the attenuation representing the ductile part. It assumes the coupling of both parts and therefore the correlation between the time series of the parameters during the normal part of the stress loading cycle. A few years before the occurrence of a major earthquake, the brittle part of the system suffers a stress overload, so it cannot maintain the normal loading processes through the brittle-ductile transition zone, provoking the decoupling of the parts and breaking the previous correlation between the analyzed time series. First, a worldwide analysis of seismicity is developed, focusing after in three regions, Colombia, Taiwan and the Iberian Peninsula. The seismicity data for the global analysis are obtained from the Centennial Catalog, while local catalogs are used for the regional study. Then, the BD hypothesis is tested in three seismic stations in Colombia and five in Taiwan; for that purpose, a selection of almost 4000 waveforms are effectively used from a total of almost 48000 of the study areas, for the coda Q quality factor analysis. The global and regional seismicity analysis allows the detection of space-time evolution trends with no easy explanation and that are attributed to external manifestations of the Earth¿s interior dynamic processes in the mantle. The detailed seismicity analysis in the 8 studied stations characterizes the brittle part. The waveforms in each station are used to determine the time series of coda Q in a time window of around 14 years. First, its space-time variation and depth dependence are studied, finding significant changes that correlate with local geotectonic structures. Besides, specific time series that characterize the ductile part are obtained. The correlation analysis and the one of the time variation of the central correlation coefficient have allowed evaluating the BD model. A new technique that allows the construction of time series of this coefficient is proposed, in a way that following and detecting the correlation/uncorrelation of the seismicity and Q time series can be done continuously. Therefore, calm and enhanced activity periods can be firmly established in the BD theory. By applying the technique to different geotectonic locations we find cases in which the behavior obeys the theory, whilst for others the validity is unclear. In the former, high values of the central correlation coefficient, above 0.6, are observed during calm periods; this value decreases drastically some time before the occurrence of a major earthquake. In the latter, the correlation is found to be, in most cases, permanently low, what can be attributed to a inexistent or poorly developed BD interaction zone, since, in most cases, these regions are ruled by complex geotectonic structures and a relatively high frequency of major earthquakes.