On May 5, 1984, a moderate earthquake with a magnitude of 4.6 occurred at the border of Kyoto and Osaka prefectures in Japan. Accompanying the earthquake, strain steps of the order of 10^-7-10^-9 were observed with extensometers at seven crustal movement observatories in the Kinki district. Observed strain steps were compared with theoretical values expected from the dislocation fault model. Some of them are found to be one or two order larger than the theoretical ones. This discrepancy can be attributed to mechanical instability of instruments or local inhomogeneities around the observation sites. Observed data are fairly well consistent with the expected values, except the extremely large ones.