Many foods are emulsions or dispersions containing lipids. The friction properties of foods are evaluated because they affect food texture and processability. Here, we evaluated the friction characteristics of 55 liquid or semisolid foods using a sinusoidal motion friction evaluation system to classify them based on friction dynamics. The contact surface was made to resemble a biological surface using agar gel, which exhibited a fractal structure, and the movement of the contact probe mimicked living movement by sinusoidal motion. The change in average friction coefficient (Δμ), static friction coefficient (Δμs) in a round trip, delay time (Δδ), and friction profile depended on the condition and rheological properties. Principal component analysis showed that all the friction parameters of Δμ, Δμs, Δδ, and the appearance ratio of the profile were involved in the principal components, Z1 and Z2 which are composite variables obtained by the contraction of many friction parameters in a principal component analysis. In addition, the foods were classified into three groups by cluster analysis using Z1 and Z2. The condition of the foods, rheological properties, and the presence or absence of lipids was the factors that defined each group.