As part of a larger joint effort by the Defense Advanced Research Project Agency and the Department of Energy to study the seismic source problem, a comprehensive reevaluation of the 1964 Salmon and 1966 Sterling nuclear explosions in dome salt was carried out. The Sterling source function originally estimated by Springer et al. (1968) conveys the impression that the cavity was badly overdriven; on reexamination this does not appear to be the case. The work of Glenn et al. (1987) on the Sterling free-field data is expanded upon, confirming that the cavity response was close to the theoretical expectation. Sterling's source function is estimated and is found to be comparable to Patterson's (1966) slightly weakened salt model. A source model for Salmon is derived from the Sterling source model and the five seismic stations that recorded both events. The new source model has a reduced displacement potential &psgr;x of about half that previously estimated. A temporary nonlinear two-wave system developed during the Salmon explosion as the compressional wave evolved from a shock wave; the separation of these two waves resulted in a high-frequency roll-off of the reduced velocity potential of ω-3. In addition, it is shown that the corner frequency is much higher and is created much closer to the cavity than the eigenfrequency. For both Salmon and Sterling the radial stresses are approximately a low-passed damped sinusoid superimposed on a small step function. The decoupling value of 72 obtained by Springer et al. (1968) is confirmed. A revision of Patterson's (1966) partial decoupling curve shows that the value for full decoupling in a shot-generated cavity would be only slightly higher. Contrary to previous studies, decoupling as a function of frequency for the surface waves is found to be the same as for the P waves. A new definition of decoupling appropriate to threshold test-ban treaty monitoring is also proposed. ¿ American Geophysical Union 1990