The microstructure of fat crystal networks is closely related to rheological properties of fat products, and thus is of particular interest to food scientists. The fractal dimensions of fat crystal networks calculated by microscopy methods such as box-counting, Db, particle-counting, Df, and mass fractal dimension, Dm, have been extensively employed to quantify the microstructure of fats. This work revealed the microstructural basis of Db, Df, and Dm in 3D space through both computer simulation and experiments on the high melting fraction of milk fat crystal networks. Similar to our previous simulation study on the fractal dimensions of fat crystal networks in 2D space, Db is sensitive to crystal size, area fraction, and not sensitive to distribution orderliness of crystals, which is the percentage of evenly distributed fat crystals within the simulated fat crystal networks. Df and Dm were not affected by any of the microstructural factors studied in the simulation.