Long-period (100 to 260 s) Love and Rayleigh waves excited by the eruption of Mount St. Helens on May 18, 1980, and recorded by ID, SRO, and ASRO stations were analyzed to determine the mechanism of the eruption. The amplitude radiation patterns of both Rayleigh and Love waves are two lobed with nodal direction in E5¿S for Rayleigh waves and in N5¿E for Love waves. These radiation patterns preclude any double-couple mechanism. The radiation pattern, the initial phase, the relatively large amplitude ratio of Love to Rayleigh waves and the existence of clear nodes in the radiation patterns of fundamental mode and higher-mode Rayleigh waves suggest that the source is represented by an almost horizontal (less than 15¿ from the horizontal) single force pointed toward S5¿W. The surface wave spectra fall off very rapidly at periods shorter than 75 s suggesting a very slow source process. Although the details of the source time history could not be determined, a smooth bell-shaped time function: f0s(t)=(1/2)f0(1-cos( (t)/(&tgr;) &pgr;)) for 0≤t≤2&tgr; and f0s(t)=0 for t≥2&tgr;, with &tgr;=75 s is considered appropriate on the basis of comparison between synthetic and observed seismograms and of the shape of the source spectrum. The peak value of the force f0 is about 1018 dynes. The tailing end of the source time function could not be resolved, and some overshoot may be added. The magnitude and the time history of the force can be explained by a northward landslide followed by a lateral blast observed at the time of the eruption. Two distinct events about 110 apart can be identified on body wave and short-period surface wave records. The first event may correspond to the earthquake which triggered the landslide and the lateral blast. The second event appears to correspond to a second large earthquake and explosion which took place about 2 minutes after the first earthquake.