Time-dependent conditional probabilities for the recurrence of large and great interplate earthquakes along the Mexican subduction zone are presented for time intervals of 5, 10, and 20 yr duration (i.e., 1986-1991, 1986-1996, and 1986-2006). At present, the central Oaxaca (97.3° to 97.7°W), Ometepec-San Marcos (98.2° to 99.5°W), and central Guerrero (100° to 101°W) segments stand out as having the highest probability for the recurrence of large and great earthquakes in the near future. Segmentation of the margin is delineated by the rupture zones of the most recent earthquakes occurring in each area. For segments of the Mexican margin with one or more known recurrence intervals, probability estimates are based on the observed average recurrence time for each segment and the lognormal probability distribution function. Use of a generic distribution function (Nishenko and Buland, 1987) allows a more stable estimate of average recurrence times than are available from a few observations. A long-term prediction time window is also defined, based on the 90 per cent confidence interval for our estimates of the recurrence time. Use of a predetermined confidence interval conveys valuable additional information as to the precision and information content of the forecast. For those segments of the margin with only one prior event, and hence, no historically observed recurrence times, repeat times are estimated by extrapolating the observed recurrence time behavior in Oaxaca, subject to the assumptions that recurrence time scales only as a function of the ratio of seismic displacement and convergence rate and that all events are characterized as simple sources (i.e., involve the rupture of a single asperity).