Recent reports suggest that not only chemical but also mechanical influences from the cellular environment could have profound effects on gene activity and could act on cell differentiation and proliferation. These mechanisms involve a tissue matrix system, which includes extracellular matrix, nuclear matrix, and cytoskeleton. Supposing that the cytoskeleton mediates mechanical transduction from the cell environment to the nucleus, significant differences in the spatial distribution of the cytoskeleton network could reflect variations in environmental signals. Hepatocytes during fetal development provide a useful model to study such variations, because a progressive establishment of intercellular contact is reported. The aim of this work is to discriminate steps in hepatocyte differentiation from fetal to adult livers, using computerized quantitative image analysis of cytokeratin (C8) immunofluorescent localization, visualized by confocal scanning laser microscopy. The filament structure is represented by the gray-scale skeleton of the digital images obtained by specially designed segmentation methods. A set of line features was investigated, including number and length of lines, orientation of lines, and the fractal dimension of the filament network. The features studied showed highly significant differences throughout liver development, with an increase of the total amount of cytokeratin filaments. We could also demonstrate a modification in the structure of the network, being more and more dense, with an increase of connecting points. Moreover, it has been shown that filaments have some orientation in fetal hepatocytes, and that directionality. Thus, significant differences in the pattern of the cytokeratin filament network according to the stages of hepatocyte differentiation have been demonstrated with objective and quantitative methods.