A statistical method for estimating the occurrence of foreshocks to larger earthquakes using local seismicity data is applied to the catalog of earthquakes from a local seismograph network near Adak, Alaska. Foreshocks are defined in a restricted sense as earthquakes preceding larger earthquakes within a critical time calculated from the mean time between successive earthquakes in the volume surrounding the target event. The threshold of completeness of the earthquake catalog is estimated as duration magnitude Md2.2. A low-magnitude cutoff on the catalog of Md2.5 was used to minimize potential bias from variations in weather and instrumentation condition. About 6 to 10 per cent of 51 main shocks of mb ≧ 4.5 in the Adak thrust zone were preceded by detectable earthquakes satisfying this statistical definition for foreshocks. As many as 14 percent of mb ≧ 5.0 earthquakes may have been preceded by detectable earthquakes satisfying this definition. An estimate of foreshock occurrence using as background the seismicity rate in the 4 months preceding larger earthquakes indicates a slightly higher rate of occurrence. The volume around the target events which showed foreshock activity is used to infer the size of a zone of preparation for larger earthquakes in the Adak area. The data are consistent with a zone of preparation whose size is a linear function of target-event magnitude. The size of the zone of preparation ranges from 20 to 45 km in radius for body-wave magnitudes 4.0 to 5.7.