Data are available on the field behavior of 39 cases involving the occurrence or nonoccurrence of liquefaction during earthquakes. The data from the 39 cases have been examined in a statistical manner to determine the relations between the field standard penetration test and the occurrence of liquefaction. The data for both relative density and accelerations are lognormally distributed, both for the data as a whole and for the data broken into the two categories of liquefaction and nonliquefaction. Statistics also indicate that the differences between these two categories involve a greater density of the soils at the sites where liquefaction did not occur rather than lower acceleration levels for the earthquakes at those sites. The statistical technique of discriminant analysis is used to locate lines separating liquefiable from nonliquefiable conditions. A line completely enveloping all observed cases of liquefaction is associated with a 90% confidence level, which is very high for the size of the data base.