The 1989 earthquake swarm beneath Mammoth Mountain, near Long Valley, California, was unique for the area because of the occurrence of low-frequency and unusual mixed-frequency events. At the time of the swarm, a computer-based earthquake detection system with near-real-time location and spectral fitting capabilities was operational at Long Valley. This system detected over 1500 earthquakes during the swarm. About 64% of the events had a high-frequency content (4 to 8 Hz), 34% had a medium-frequency content (<4 Hz only), and 2% were low- or mixed-frequency events (∼2 Hz). Unusual events detected by the computer system during the 6-month swarm were 1) spasmodic tremor (14 strong bursts); 2) low-frequency events (30); and 3) low-frequency events with high-frequency riders, or mixed-frequency events (4 of the 30 low-frequency events). Filtering of mixed-frequency event waveforms suggests that they are a low-frequency event triggered by an earthquake. Mixed-frequency events had a limited source depth of 4 to 6 km while most low-frequency events were shallow. However, earthquakes with lower than average frequencies had source depths of 1 to 9 km, which include most of the depth range of the swarm earthquakes. Based on a comparison with tremor models by Chouet (1992), fluid resonance is a satisfactory explanation for the observed mixed-frequency events, although not the only possible explanation. Estimated resonant cavity size is 100 to 150 m for the Mammoth Mountain swarm. Geodetic and geothermal as well as seismic data suggest that an intrusion of water or magma probably occurred beneath Mammoth Mountain during this swarm. Finally, the computer-based detection and data-processing system used in this study demonstrated the value of spectral fitting capabilities for monitoring volcanic areas.