In November 1995, we installed five, three-component broadband seismometers and electronic tiltmeters around the circumference of Arenal Volcano, a young stratovolcano in Costa Rica that exhibits strombolian activity. With the addition of two continuous-recording GPS receivers deployed in May 1995, these instruments provide continuous monitoring of seismicity and ground deformation at an active volcano over a very wide bandwidth. In addition, during April–May 1997, we deployed a small, linear array of co-located three-component seismometers and broadband microphones. This paper presents a comprehensive analysis of all the seismic and acoustic data collected thus far. Seismic signals are primarily of two types: (1) long-period (1–3 Hz) transients associated with summit explosions; and (2) harmonic tremor that contains regularly spaced spectral peaks (0.9, 1.8, 2.7, 3.6, 4.5, 5.4, 6.3, and 7.1 Hz) and lasts up to several hours. The explosion signals appear to originate in a small volume that is located at shallow depth beneath the vent and does not migrate with time. No unambiguous long-period seismic signals (T>5s)associated with volcanic processes at Arenal have been observed during the three-year deployment period. The spectra of summit explosions show distinct signatures at each site, suggesting significant path and/or site modification of the waveforms. In contrast, the harmonic tremor signals show no variation in the frequency content at the five sites, nor on the three seismic components at each site (Hagerty et al., 1997). This, and the fact that harmonic tremor is recorded in the acoustic channels as well, demonstrates that harmonic tremor is not a seismic propagation effect and that pressure disturbances propagate within the magma–gas mixture inside of volcanic conduits. These pressure waves are sensitive to the flow velocity and to small changes in the gas content of the magma–gas mixture. Observations and synthetic tests are presented that challenge the notion that harmonic tremor is a superposition of repeated gas explosions at shallow depth. We propose that equilibrium degassing of the melt creates a stable, stratified magma column where the void fraction increases with decreasing depth and that disruption of this equilibrium stratification is responsible for observed variations in the seismic efficiency of explosions.