“Predictions” of imminent volcanic eruptions are more successful than are “forecasts” of impending activity if we use the terminology defined by Swanson in the accompanying article. The implication is that, despite occasional intervals of periodically recurring eruptions, the long term (months to years) activity of volcanoes is stochastic. Once magma rises near the surface, however, and initiates measurable phenomena (harmonic tremor, inflation, increased fuming, etc.), a volcano appears to be locked into a nonreversible process leading to an eruption weeks to hours hence. Each type of forewarning is valuable, and USGS volcanologists have demonstrated that the basic monitoring and prediction techniques developed for effusive eruptions in Hawaii are transferable to explosive activity in the Cascades. But longer term forecasts, as pointed out by Swanson, are still largely unreliable.