Extremely hazardous volcanic dome collapse and pyroclastic flows can occur without warning and can be triggered by intense rainfall. Rainfall data collected on Soufriere Hills Volcano, Montserrat between 1998 and 2006 was analysed to assess the impact on primary volcanic activity, defined here as pyroclastic flows, dome collapse and explosions and to explore the utility of meteorological data as a predictive tool for active volcanoes. A statistical analysis of daily rainfall totals and volcanic activity, showed that if greater than 20 mm of rainfall fell on a particular day the probability of a dome collapse occurring on that day increased by a factor of 6.2 from 1.0% to 6.1%, similarly the probability of observing pyroclastic' flows and explosions increased by factors of 2.0 and 4.3 respectively. These statistically significant links increased as the rainfall threshold increased, and in some cases activity was observed up to 48 hrs after rainfall. The state of the volcano is important and an absence of a significant volume of material above the conduit removes the possibility that rainfall could trigger a collapse. An analysis of a network of 1 minute resolution rainfall and seismic data for the period 1 January 2001 - 31 December 2003 showed a correlation, statistically significant to the 99% level was found between rainfall and heightened activity; rockfalls, long period earthquakes, long period rockfalls and hybrid earthquakes. There were three prominent peaks in magnitUde, following a rainfall event of 5 mm hr-I at time lags of approximately 5, 14 and 25 hours. Over 50% of the heavy rainfall days were associated with synoptic scale weather systems which are potentially well predicted by current forecast models. However the remaining days were associated with localized and essentially unpredictable systems. There was significant variability between raingauges reflecting topographic variations and inherent variability within weather systems, hence any monitoring program is recommended to use a network rather than a single gauge. Incorporation of weather forecasting into hazard monitoring programs would help to improve the prediction of hazardous collapses.