A broadly based report on seismic hazards in southern California (WGCEP, 1995) concluded that the predicted seismicity exceeds that observed since 1850; a subsequent independent analysis argued that infrequent huge (M > 8) earthquakes are needed to explain the low rate of large earthquakes (Jackson, 1996). Frequency-magnitude relationships and earthquake reporting suggest that the 1903 to 1997 catalog we present here, with a b-value of 1.0 and a rate of M ≧ 6 shocks of 0.42 to 0.49 yr−1, is nearly complete. In contrast, the 1850 to 1994 catalog used by WGCEP is incomplete before the turn of the century, and thus its reported seismicity rate of 0.32 M ≧ 6 shocks yr−1 is too low. Principally because the WGCEP (1995) model results in b-values of up to 4.0 for regions of lesser and blind faults, the rate of M ≧ 6 shocks off the San Andreas system predicted by the WGCEP (1995) model is three times greater than that observed in this century. Because they obtained b = 0.4 for M < 7.3 and b = 2.2 for M ≧ 7.3 on major faults, their expected rate of M ≧ 7 San Andreas shocks is twice as high as observed. Thus, part of the seismicity and moment discrepancy identified by WGCEP was caused by use of an incomplete catalog, and part was caused by inappropriate b-values. We obtain a southern California moment release rate of 8 to 12 × 1018 N-m yr−1, which cannot be distinguished from the moment release estimated by fault slip, or the moment accumulation inferred from plate motions or geodetically measured shear strain. We thus find no evidence for a moment deficit, significant aseismic moment release, or for rare M > 8 earthquakes off the San Andreas fault system. Finally, the number of M ≧ 6 earthquakes per decade does not depart significantly from a Poisson process during this century, and thus we find no evidence that the rate of seismicity is increasing, now or at any other time since 1900.