By using rank-order statistics, it is possible to predict the likely s
cale of future ''extreme'' volcanic events (eruptions larger than a gi
ven threshold size, or repose periods longer than a given time) on bot
h a local and global scale. When ranked by diminishing size, the large
st and most devastating volcanic eruptions in terms of volume, or cons
equent fatalities, ran be described by power-law functions. In turn, t
his approach permits projection of, for example, (1) the Likely size o
f the next volcanic eruption larger than the A.D. 1815 eruption of Tam
bora (similar to 90 km(3) of magma); (2) the size of the next eruption
of the Taupo volcano, New Zealand, that is larger than the eruption o
f A.D. 186 (similar to 82 km(3) of tephra); and (3) the likely duratio
n of the current repose period of the Taupo volcano (4000 +/- 1200 yea
rs). This approach can also be employed on shorter time scales, for ex
ample to provide quantitative criteria to assess when a period of acti
vity has ceased. With the establishment of procedures for making proje
ctions such as these, volcano hazard assessment may be placed onto a m
ore rigorous and quantitative basis.