We calculate the evolution of stresses in southern California, extending the study of Deng and Sykes <1997> by increasing from 6 to 36 the number of earthquakes for which coseismic changes in stress are computed and by expanding from M≥6 to M≥1.8 the range of magnitudes M of events whose focal mechanism solutions are examined in the context of the evolving stress field. The cumulative stress on a given date is calculated with respect to an arbitrary zero baseline just before the 1812 Wrightwood earthquake. By taking into account the long-term stress loading associated with 98 fault segments and coseismic stress changes for 36 significant earthquakes, our calculations indicate that more than 85% of M≥5 earthquakes from 1932--1995 occurred in regions of positive change in Coulomb failure function (ΔCFF). Most of the remaining about 15% earthquakes that occurred in areas of negative ΔCFF fall very close to boundaries between positive and negative ΔCFF, some of which are sensitive to the less well controlled slip distributions of the earliest historic events. Calculations also show that from 1981 until just before the 1992 Landers earthquake more than 85% of small-(M≥3) and micro-size (M≥1.8) shocks in the Seeber and Armbruster <1995> catalog with mechanisms involving either NW trending right-lateral or NE trending left-lateral strike-slip faulting occurred in regions of positive ΔCFF. The ratio of encouraged to all small- and micro-size events reaches a high value of about 88% if an apparent coefficient of friction μ between 0.0 and 0.6 is used. The highest percentage of earthquakes occurred in areas where stress is about 1 MPa above the 1812 baseline. Most (66%) events occurred in regions of ΔCFF between 0.0 and 2.0 MPa. The upper limit indicates that the approximate range of stress variation in the earthquake cycle is of the order of 2.0 MPa. The fact that the locations of most moderate-, small-, and micro-size earthquakes are still related to stress changes remaining from large historical events might be used to constrain slip distribution of some of those earthquakes and to constrain the locations of future significant events.¿ 1997 American Geophysical Union