Podoplanin, a type I transmembrane glycoprotein with an effect of platelet aggregation, has been reported to be one of the possible prognostic factors of oral squamous cell carcinoma (OSCC). However, the biological significance of podoplanin is largely unclear. The aim of this study was to develop a practical model for the prediction of prognosis using the grade of podoplanin expression, and also to evaluate the biological function of podoplanin. Eighty-two specimens of patients with previously untreated OSCC, who underwent either biopsy or surgery, were histopathologically and immunohistochemically analyzed. These 82 cases were composed of 66 well-differentiated, 10 moderately differentiated and 6 poorly differentiated OSCC. Podoplanin was successfully immunostained in 78 specimens, and was detected in most cases, but the frequency of positive cells varied. The prognosis of patients with more than 50 % podoplanin-positive tumor cells was significantly poorer than that of the other patients. Multivariate hazards regression analysis suggested that a linear combination of covariates, OSCC patients with more or less than 50 % podoplanin expression, age of more or less than 70 years old, mode of invasion and T3, T4 or T2 versus T1 of the UICC T-stage classification was the most effective model for evaluating the prognosis of OSCC patients. Additionally, podoplanin expression had a significant relationship to UICC clinical stage and the expression of Ki-67. An effective regression model using podoplanin expression was developed for evaluating the prognosis of OSCC and the biological significance of podoplanin was suggested to be associated with the growth and/or progression of OSCC.