A method of analysing jaw movements was developed by converting jaw displacements into a matrix of frequency distributions. Low frequencies were first filtered out; then quartiles and modes were used to describe the remaining core. Eight different types of measures (variables) were developed to describe the proportions and density of a core of movement. These variables were repeatable at successive recording sessions, yet accurate enough to reveal differences in chewing patterns between right- and left-sided chewing. All eight variables of jaw displacement were required to discriminate between the chewing sides of 15 subjects. A statistical model was developed to express the principal components of jaw movement. The first component consisted of variables that expressed the shape of the chewing cycle; the second, its distribution about the midline; the third, the prevalence of a bimodal pathway in the sagittal plane. There was a wide variation in patterns of chewing movements, which appear to be consistent with normal masticatory function. A multivariate model appears to be required to describe chewing movements.