Abstract.The state of tectonic stress in the epicentral area of the 17 January 1994, Northridge earthquake (Mw 6.7) is investigated by applying a stress inversion method to P-wave polarity data from earthquakes in Northridge from July 1981 to January 1994 and from the Northridge aftershocks during January 1994 to December 1995. A 3-D crustal model is used to trace the rays taking off from the hypocenter, which reduced the effects of large structural heterogeneities on the determination of the stress tensor. We found significant temporal changes of stress orientations induced by the Northridge earthquake. The principal pressure (P) axis is oriented N32øE from 1981 to June 1992, and N30øE from 28 June 1992 to 16 January 1994, suggesting that the stress field in Northridge was not affected by the 1992 Landers earthquake. During two weeks following the Northridge mainshock, the P-axis is oriented N13øE, which is a significant (17 ø) change from that before the earthquake (N30øE). Between February 1994 and August 1995 the P-axis orientation changes from N18øE to N26øE, and finally ends up at N34øE by the end of 1995, which is close to that before the Northridge earthquake. These results suggest that the stresses to. rated coseismically, then rotated more slowly back to their original orientation. The aftershocks caused by the mainshock changed the stress distribution in the crust, which showed up as a regional stress change. The stress recovery appears to have completed within two years after the mainshock, which is very short compared to the time scale of the earthquake cycle.vided an excellent opportunity for investigating in more detail the possible temporal and spatial stress changes in the epicentral area before and after a major earthquake. The Northridge area is covered by the dense