It has been suggested that shear fractures initiate parallel to lines of no finite elongation (LNFE). This idea is attractive, because it opens up the possibility of using simple kinematic models (e.g. trishear) to predict the orientations of shear fractures in subsurface reservoirs. We test the hypothesis that cataclastic deformation bands that cut unconsolidated late Pleistocene sands at McKinleyville, California developed parallel to LNFE. Here, the deformation bands form two distinct sets that dip shallowly towards the north-northeast and south-southwest. The acute dihedral angle between the two sets of deformation bands is c. 47 and is bisected by the sub-horizontal, north-northeast directed instantaneous and finite shortening directions. Two-dimensional trishear models of fault propagation folding above the initially buried tip of the McKinleyville Fault predict two sets of LNFE that plunge steeply and shallowly to the south and north, respectively. These predictions are inconsistent with the observed cataclastic deformation band orientations and suggest that the deformation bands did not form parallel to these LNFE. Three-dimensional strain modelling using a Mohr construction predicts that the cataclastic deformation bands at McKinleyville could instead have formed parallel to zero extension directions, provided deformation band localisation was accompanied by a transient volume increase of ca. 4%. The inference of a small transient volume increase is consistent with previous experimental and geological observations. The hypothesis that the two sets of cataclastic deformation bands at McKinleyville formed parallel to zero extension directions is therefore tenable, but further research is required to quantify transient volume changes during deformation band localisation.