On May 10, 2003, the first historical eruption of Anatahan volcano in the western Pacific Mariana Islands was fortuitously recorded by a broadband seismograph installed on the island only 4 days prior to the eruption. This station, located 7 km WNW of the active crater, together with another broadband seismograph on Sarigan Island 45 km to the north, continued to operate throughout the 2-month period of major eruptive activity in May and June and throughout the majority of the following year. In June 2003, the Saipan Emergency Management Office and the US Geological Survey installed two telemetered high-gain short-period seismic stations to monitor the activity in real-time. The only earthquakes detected in the 4-day period from the initial seismograph installation until 6 h prior to the eruption occurred approximately 20 km to the northeast of the island on May 8. The first volcano-tectonic (VT) event located near the volcano occurred at 01:53 GMT on May 10. The number of events per hour then increased dramatically and a period of about 80 discrete earthquakes per hour commenced at about 06:20 GMT, immediately prior to the estimated eruption time of 07:30 from the Volcanic Ash Advisory Center. A long-period tilt signal recorded on the horizontal components of the broadband seismograph, indicating upward movement of the crater region, also commenced at about 06:20. Inflation continued until 09:30, when the direction of tilt reversed. Deflation continued until 17:50, coinciding with a reduction in the number of VT events. The larger VT events were located with a linearized least-squares location algorithm. Magnitudes of located VT events on May 10 ranged from 2.0 to 3.2, but a period of larger VT events were recorded on May 11, with the largest M 4.2. After about 36 h of intense earthquake activity, the number of discrete VT events declined and was replaced by nearly continuous volcanic tremor for the next 6 weeks. Differing types of very long-period events may suggest complex non-destructive magmatic source mechanisms, persisting dominantly throughout the first 10 days of the initial eruption. Visual reports indicate that a small craggy dome extruded sometime between May 20 and June 5. From analogy with other volcanic dome extrusions, we believe the dome probably extruded just as the tremor amplitude decreased dramatically about May 24. This dome was then destroyed between June 13 and 14. Reduced displacement of the co-eruption tremor is estimated as approximately 40-80 cm2, suggesting an eruption with a Volcanic Explosivity Index of about 3. 2005 Elsevier B.V. All rights reserved.