The seismic potential for 96 segments of simple plate boundaries around the circum-Pacific region is presented in terms of the conditional probability for the occurrence of either large or great interplate earthquakes during the next 5, 10, and 20 years (i.e., 1989-1994, 1989-1999 and 1989-2009). This study represents the first probabilistic summary of seismic potential on this scale, and involves the comparison of plate boundary segments that exhibit varying recurrence times, magnitudes, and tectonic regimes. Presenting these data in a probabilistic framework provides a basis for the uniform comparison of seismic hazard between these differing fault segments, as well as accounting for individual variations in recurrence time along a specific fault segment, and uncertainties in the determination of the average recurrence time. The definition of specific segments along simple plate boundaries relies on the mapping of earthquake rupture zones as defined by the aftershock distributions of prior large and great earthquakes, and historic descriptions of felt intensities and damage areas. The 96 segments are chosen to represent areas likely to be ruptured by "characteristic" earthquakes of a specified size or magnitude. The term characteristic implies repeated breakage of a plate boundary segment by large or great earthquakes whose source dimensions are similar from cycle to cycle. This definition does not exclude the possibility that occasionally adjacent characteristic earthquake segments may break together in a single, larger event. Conversely, a segment may also break in a series of smaller ruptures. Estimates of recurrence times and conditional probabilities for characteristic earthquakes along segments of simple plate boundaries are based on 1) the historic and instrumental record of large and great earthquake occurrence; 2) paleoseismic evidence of recurrence from radiometric dating of Holocene features produced by earthquakes; 3) direct calculations of recurrence time from the size of the most recent characteristic event and the long-term rates of plate motion assuming the validity of the time-predictable model for earthquake recurrence; and 4) the application of a lognormal distribution for the recurrence times of large and great earthquakes. Time-dependent estimates of seismic potential are based on a physical model of earthquake occurrence which assumes that the probability for an earthquake is low immediately following the occurrence of a characteristic earthquake and increases with time as the stress on the fault segment recovers the stress drop of the event. This study updates earlier work on seismic gaps by explicitly including both recurrence time information and the temporal proximity to the next event as factors in describing earthquake hazards. Currently, 11 out of 96 regions have a high (i.e., ???50%) probability of recurrence during the next 10 years and are characterized by either fairly short (i.e., less than 30-40 years) recurrence times or long elapsed times relative to the average recurrence time. The majority of these segments are located in the southwest Pacific (Vanuatu, New Guinea, and Tonga). When a longer time window is considered (e.g., 20 years or 1989-2009), 30 out of 96 regions have a high potential. Many of these regions are located near areas of high population density. These determinations do not preclude rupture of other fault segments, with less than a 50% chance in 10 or 20 years, or large and great earthquakes in areas we have not studied in detail. While this study has summarized the seismic potential for a large number of regions around the circum-Pacific, there are still a number of geographic and seismotectonic regions that need to be considered, including Indonesia, the Philippines, New Zealand, and the countries that surround the Caribbean basin. ?? 1991 Birkha??user Verlag.