We estimated the energy radiated by earthquakes in southern California using on-scale very broadband recordings from TERRAscope. The method we used involves time integration of the squared ground-motion velocity and empirical determination of the distance attenuation function and the station corrections. The time integral is typically taken over a duration of 2 min after the P-wave arrival. The attenuation curve for the energy integral we obtained is given by q(r) = cr−nexp(−kr)(r2 = Δ2 + href2) with c = 0.49710, n = 1.0322, k = 0.0035 km−1, and href = 8 km, where Δ is the epicentral distance. A similar method was used to determine ML using TERRAscope data. The station corrections for ML are determined such that the ML values determined from TERRAscope agree with those from the traditional optical Wood-Anderson seismographs. For 1.5 < ML < 6.0, a linear relationship log ES = 1.96 ML + 9.05 (ES in ergs) was obtained. However, for events with ML > 6.5, ML saturates. The ratio ES/M0 (M0: seismic moment), a measure of the average stress drop, for six earthquakes, the 1989 Montebello earthquake (ML = 4.6), the 1989 Pasadena earthquake (ML = 4.9), the 1990 Upland earthquake (ML = 5.2), the 1991 Sierra Madre earthquake (ML = 5.8), the 1992 Joshua Tree earthquake (ML = 6.1), and the 1992 Landers earthquake (Mw = 7.3), are about 10 times larger than those of the others that include the aftershocks of the 1987 Whittier Narrows earthquake, the Sierra Madre earthquake, the Joshua Tree earthquake, and the two earthquakes on the San Jacinto fault. The difference in the stress drop between the mainshock and their large aftershocks may be similar to that between earthquakes on a fault with long and short repeat times. The aftershocks, which occurred on the fault plane where the mainshock slippage occurred, had a very short time to heal, hence a low stress drop. The repeat time of the major earthquakes on the frontal fault systems in the Transverse Ranges in southern California is believed to be very long, a few thousand years. Hence, the events in the Transverse Ranges may have higher stress drops than those of the events occurring on faults with shorter repeat times, such as the San Andreas fault and the San Jacinto fault. The observation that very high stress-drop events occur in the Transverse Ranges and the Los Angeles Basin has important implications for the regional seismic potential. The occurrence of these high stress-drop events near the bottom of the seismogenic zone strongly suggests that these fault systems are capable of supporting high stress that will eventually be released in major seismic events. Characterization of earthquakes in terms of the ES/M0 ratio using broadband data will help delineate the spatial distribution of seismogenic stresses in the Los Angeles basin and the Transverse Ranges.