The Imperial Valley earthquake was a multiple sequence (Trifunac and Brune, 1970) with at least four events occurring during the main energy release. These four events, recorded on the strong-motion seismograph in El Centro, located about 10 km NW from the instrumentally determined epicenter, and nine aftershocks recorded in the next 5 min are re-examined in this paper to test an approximate source theory (Brune, 1970). This theory predicts the shape of the body-wave spectra in terms of the seismic moment and stress drop. By fitting theoretical spectra to the spectra calculated from the strong-motion accelerogram, moment and stress drop can be estimated for each of the multiple events. Inasmuch as the average displacements at the fault and the source dimensions can be derived from the known moment and stress drop, the pattern of average displacements along the fault was computed from the instrumental records. A test of the theory, then, consists of comparing the fault displacements derived from seismograms with the fault displacements observed at the surface (Buwalda, unpublished field notes). For the Imperial Valley earthquake, agreement between these two independent methods of measurement is good, suggesting that the above theory is an adequate first approximation for the spectra of body waves. The stress drop variations along the fault, inferred also from the above theory, indicate two areas of major stress concentration located near the northwestern and southeastern ends of the dislocation. The stress drops for various events varied from about ten to several hundred bars.