On 18 July 1995, after more than three years of irregularly increasing seismicity, phreatic explosions opened a new vent on Soufrière Hills Volcano, about 4 km east of the capital city of Plymouth, Montserrat. By early August 1995, the volcano was monitored by a nine-station seismic network, three telemetered electronic tiltmeters, and daily correlation spectroscopy (COSPEC) flights to measure SO2 emission rates and to observe vent areas. Seismicity and SO2 emission rates implicated magma intrusion as the cause of the seismic unrest. Strong evidence of magma ascent to shallow levels, however, did not appear until late September 1995, when increasing numbers of hybrid events heralded the formation of a small dome and spine. We infer from the data that intrusion of a small volume of magma occurred in July, but stalled on ascent. Degassing of the stalled magma formed a carapace that thickened with time. We suggest that either volatile build-up beneath the degassed carapace, or aseismic intrusion of fresh material, finally forced the stalled magma to the surface in late September 1995. A similar cycle of activity appears to have occurred during the second phreatic phase between late September and mid-November 1995.