Twenty ground stations distributed over a 40-square-mile area in Pasadena recorded strong ground shaking during the San Fernando earthquake of February 9, 1971. Relative responses at 10 of these same stations as measured for small earthquakes by standard Wood-Anderson torsion seismometers are available for comparison from a study made by Gutenberg in the 1950's. Frequency spectra of strong ground motions as calculated for four sites having time-recording accelerographs assist in the interpretation of seismoscope results at the other stations. Attempts to correlate local distributions with known features of local geology such as thickness of alluvium, distance from known faults, etc., indicate that no single feature plays a dominant role in the resulting patterns. The implications of such complicated distributions for the preparation of seismic risk maps are discussed, and it is concluded that it would not be possible in the present state of knowledge to assess meaningful variations in the seismic risk throughout the Pasadena area.