The 28 June 1992 Big Bear earthquake occurred at 15:05:21 GMT and is considered to be an aftershock of the earlier Mw = 7.3 Landers earthquake. From overall aftershock locations and long-period focal studies, rupture is generally assumed to have propagated northeast. No surface rupture was found, however, and the mainshock locations determined from both strong motion and TERRAscope data are mutually consistent and do not lie on the assumed fault plane. Further, directivity analysis of records from the TERRAscope array suggests significant short- and long-period energy propagating northwest along the presumed antithetic fault plane. This observation is supported by significant early aftershocks distributed along both the presumed rupture plane and the antithetic plane to the northwest. An empirical Green's function (eGf) approach using both the Mw = 5.2, 28 June 1992 14:43 GMT foreshock and the Mw = 5.0 17 August 1992 aftershock produces consistent results and suggests that the Big Bear event comprised at least two substantial subevents. From the eGf results, we infer that the second and possibly a third subevent occurred on the presumed (northeast striking) mainshock rupture surface, but that significant moment release occurred on the antithetic northwest striking surface. We present results from line-source fault modeling of broadband displacement recordings of the Big Bear mainshock, which indicate that a two-fault event is necessary to produce the observed waveforms. The limitations imposed by the mainshock location and directivity analysis require that the initial rupture be towards the northwest, striking 320°. This was followed approximately 4 sec later by bilateral rupture along a northeast-southwest fault that strikes 50° east of north.