Following two moderate earthquakes, reports of the behavior of animals prior to the earthquakes were gathered using a standardized interview schedule. This schedule was developed to maximize the reliability and validity of the reports through the application of accepted social science methodology. Interviews with 50 households near Willits, California, produced 17 reports of unusual animal behavior prior to that earthquake. Only one of thirty-five interviewees who experienced a similar earthquake near Ovando, Montana, reported unusual animal behavior prior to the earthquake. This difference in the frequency of positive reports and the content of the positive reports from Willits support the inference that a number of animals at Willits were responding to physical precursors which were absent at Ovando. The behavior reported at Willits was unusual in the sense that there was no immediate explanation for it, but it was not bizarre. On the contrary, it was always behavior typical of that species when motivated by generalized anxiety. Unusual behavior was often reported in only one or a few animals at a particular location so that even at Willits most animals were described as having behaved normally.