In diabetic patients, dry eye and other ocular surface diseases occur more often than in healthy subjects. The aim of this study was to analyze the tear protein patterns of patients suffering from diabetes mellitus type II (dia) and to compare them to the patterns of healthy volunteers (ctrl). Tear proteins of nonstimulated tears of 20 patients (ctrl n=10, dia n=10) were separated using two-dimensional electrophoretic techniques. The protein patterns of each group were analyzed by a multivariate analysis of discriminance. Furthermore, for all spots of each gel, a 50 x 50 variables pH/M-r (molecular weight) array was generated and subsequently analyzed by a multivariate analysis of discriminance. Additionally, an artificial neural network was trained using the matrix data as input and a sensitivity analysis was performed to figure out, which spots were the most important to differentiate between the tear protein patterns. In both groups a complex staining pattern could be obtained. In diabetic patients significantly more spots were detected compared to the control group (P <0.02). The analysis of discriminance found a highly significant difference between dia and ctrl (P <0.00001). Using the matrix data, the analysis of discriminance showed a significant difference between the two groups, too (P <0.0001). The sensitivity analysis by means of the artificial neural network revealed several spots that were more expressed or more frequently present in the diabetic group. Our findings reveal that the composition of tear proteins of diabetic patients is different from that of healthy subjects. The use of the two-dimensional electrophoretic technique could give more insight into the diabetic-related changes in the tear film composition.