Oral squamous cell carcinoma (OSCC) is a highly aggressive cancer that is characterized by a high rate of invasion and destruction to the surrounding tissues, with patients showing poor 5-year survival rate. A pre-malignant stage of cellular atypia and loss of stratification within the epithelium (dysplasia) occurs prior to the establishment of OSCC, which is manifested as white (leukoplakia) or red (erythroplakia) lesions. Treatment of dysplasia/OSCC involves surgical intervention with the removal of an adequate safety margin. However, high grade dysplasia and OSCC exhibit high recurrence rates. To date the only method used to diagnose oral epithelial dysplasia (OED) and OSCC is haematoxylin and eosin staining of biopsies and examination by an experienced pathologist. Furthermore, the mechanisms of dysplasia formation, transition to OSCC, and high recurrence rates are little understood. Recent data from the Ward lab suggests that the cell surface tumour suppressor protein E-cadherin plays an important role in regulating many cellular functions in epithelial cells. Loss of E-cadherin in carcinomas has been well studied and is linked with tumour invasion, metastasis and poorer patient outcomes. In this thesis, I have investigated expression of E-cadherin in low grade (LG) and high grade (HG) dysplasia and T1 and T4 OSCC patient biopsies to determine whether loss of this protein occurs prior to tumour cell invasion. Furthermore, microarray data analysis identified Epithelial Membrane Protein-1 (EMP-1) as a putative early marker of tumorigenesis, with alterations of N-cadherin, CD44 and 5T4 oncofoetal antigen expression during tumorigenesis inferred from our studies in embryonic stem cells. Immunofluorescence microscopy (IFM) analysis revealed that normal oral epithelium exhibits cell surface E-cadherin, EMP-1 and 5T4 expression but generally lacks N-cadherin and CD44 reactivity. The statistically significant loss of both E-cadherin and EMP-1 was observed in low and high grade dysplastic tissue and OSCC biopsies (p < 0.001). 5T4 expression was decreased in HG dysplasia (p < 0.001), suggesting this may be a useful assay for discriminating between LG and HG dysplastic tissues. N-cadherin or CD44 did not show any statistical significance in expression between normal, LG/HG dysplasia and T1/T4 OSCC, although there was a trend for increased N-cadherin expression in T4 OSCC. Significantly, Ecadherin expression was decreased or absent from surgical margins of LG/HG dysplasia and T1 OSCC, and EMP-1 and 5T4 were absent from the margins of LG and HG dysplastic tissue, indicating that these margins are abnormal. Therefore, loss of Ecadherin and EMP-1 are early events associated with dysplastic tissue and may be a useful method to confirm the removal of sufficient surgical safety margin. The OSCC cell line BICR56 was assessed for marker expression and exhibited a phenotype consistent with normal oral epithelium. Inhibition of E-cadherin protein in BICR56 cells using a neutralising antibody led to decreased cell surface localisation of 5T4 antigens, suggesting that E-cadherin expression is critical for maintenance of a normal epithelial phenotype. In conclusion, both E-cadherin and EMP-1 are useful markers for detecting early cancerous changes and 5T4 expression can discriminate between LG and HG dysplasia. This study has also shown significant abnormalities within the surgical safety margins of LG/HG dysplasia and T1 OSCC biopsies, which requires further investigation, and may explain the high recurrence rates seen in such patients.