A series of laboratory experiments has been performed to study the decoupling of explosions by placing them in rubble-filled cavities. The source conditions consisted of a tamped source (for reference), a spherical cavity, and two sizes of cylindrical cavity located at the center of a cylindrical core of granite. The rubble was well-graded granite aggregate compressed to provide a porosity of 20% (one test at 30%). Measurements consisted of particle velocities at several ranges from which displacements, reduced velocity and displacement potentials, spectra, and radiated kinetic energy were obtained. On the basis of the results of these small-scale experiments, conclusions are drawn concerning the extent of yield underestimation associated with the rubble-filled cavities and the dependence of the estimation on the decoupling criterion selected. Underestimation can amount to large factors, and a cavity test is identifiable by means of spectra in which the cavity natural frequency is evident. ¿ American Geophysical Union 1993