Using data recorded on 7- and 19-element subarrays at TFO, we compare the relative abilities of the beam and a mixed-signal processor to attenuate P-wave codas. The P-wave signals from various earthquakes are time-shifted to simulate arrivals from different azimuths and 60° distance. These signals are assumed to mask an explosion detonated at Semipalatinsk, about 96° distant from TFO. The analysis consists of determining how much of signal 1 (the earthquake) leaks into our estimate of signal 2 (the event being masked). Because the beam and mixed-signal processor are linear processors, signal 2 need not actually be present; we need only determine how much of signal 1 is present in our estimate of signal 2. The coda attenuation capability of the mixed-signal processor is found to exceed that of the beam. Up to 14-db improvement over the beam is obtained, although improvement is generally on the order of 3 to 5 db for both subarrays. A nominal value of 18 db is representative of the maximum coda attenuation obtained using the mixed-signal processor and the 19-element subarray; for the 7-element subarray, maximum attenuation is roughly 14 db. Coda attenuation obtained using the 7-element subarray and the mixed-signal processor is comparable to that obtained using the 19-element subarray and the beam. Preliminary results suggest that when the difference in the ray parameter vectors is small, a large-aperture array and the mixed-signal processor are required for significant (∼9 db) coda attenuation.