Seismology is a useful tool to gain a better understanding of volcanic unrest in real time as it unfolds. The generation of seismic signals in a volcanic environment has been linked to a number of different physical processes occurring at depth, including fracturing of the volcanic edifice (producing high frequency seismicity) and movement of magmatic fluids (producing low frequency seismicity). Further classification of seismic signals according to their waveform similarity, in addition to their frequency content, allows greater detail in temporal and spatial changes of seismicity to be detected. At Soufriere Hills volcano, Montserrat, one of the target volcanoes of the VUELCO project, families of similar waveforms provided valuable insight into evaluating the significance of ongoing unrest. In June 1997 over 6000 more events were able to be identified over a 5 day period of interest (22 to 25 June) by using families of seismic events, rather than a standard amplitude-based detection algorithm. In total, 11 families were identified, with the events clustering into a number of swarms, suggesting a repeating and non destructive cyclic source mechanism. Since each family is believed to represent a distinct source location and mechanism, identifying 11 coexisting families reflects the complex diversity of physical processes which act simultaneous at this volcano. In July 2003, conditions at the volcano had clearly changed since only one family of seismicity was identified. The source location of this family appeared to shift with time from 8 July (when no events from the family were identified) to 12 July (where most events had a cross correlation coefficient over 0.9). In addition, the use of families appears to greatly aid hindsight forecasting attempts for the large scale dome collapses of 1997 and 2003 using the Failure Forecast Method. Knowledge of the temporal and spatial extent of seismicity during periods of unrest, its source mechanism and its relationship to physical processes at depth is essential for decision and policy makers for risk mitigation. However, the source mechanisms of such volcanic seismicity is still much debated and appears to often be misinterpreted because of compromising assumptions used in the numerical modelling of inverting such sources. Use of a spatially extended source such as a ring fault structure, rather than a single point for determining the origin of low frequency seismicity, is now thought to be more realistic for the mechanism of such events since it more accurately represents the movement of magma through a conduit. However, use of this spatially extended source instead of a simple single point results in a large underestimation of slip from P-wave amplitudes, which may lead to an underestimation in magma ascent rates, with large consequences for eruption forecasting. Additionally, the P-wave radiation patterns exhibited by these two mechanisms are remarkably similar, and can only be distinguished if the small radial radiation lobes can be determined. In a volcanic environment this is extremely difficult due to large uncertainties in earthquake source depth locations, and the implementation of small aperture seismic networks.