Objectives. p63, a p53 homologue, may be associated with tumorigenesis in epithelial tissues through its inhibition of p53 transactivation functions. We sought to determine the pattern and levels of p63 expression in oral dysplasias and carcinomas using standard immunohistochemical staining. We also assessed and compared expression of p53 and a cell proliferation marker in these lesions. Study design. This retrospective cross-sectional survey (n = 67) included hyperkeratosis (10), mild dysplasia (9), moderate dysplasia (11), severe dysplasia/in situ carcinoma (10), squamous cell carcinoma (SCC) (22 [9 well differentiated, 7 moderately differentiated, 6 poor differentiated]), and normal mucosa (5). Serial sections were stained immunohistochemically with antibodies to p63 (4A4 recognizing all p63 isotypes), p53 (DO-7), and Ki-67 (MIB-1) proteins. In preinvasive lesions, both the percentage of positive cells and staining patterns (negative, basal, suprabasal) were assessed. In oral SCCs, the percentage of positive cells was assessed. Statistical analysis was done using the Tukey-Kramer multiple comparisons test. Results. A suprabasal p63 staining pattern was evident in keratinocyte nuclei in the entire range of noninvasive lesions studied, including normal mucosa. Most nuclei in invasive SCCs stained positive. When all grades of dysplasia were combined, the percent of p63 positive cells was significantly greater than hyperkeratosis (P < .01), and well-differentiated SCC (P < .001). Moderately differentiated SCC had statistically significant more positive cells than well-differentiated SSC (P < .01). Comparison of serial sections showed different p63 staining patterns compared to p53 or Ki-67 staining patterns. Conclusions. We conclude that p63 is expressed in oral carcinomas and dysplasias, as determined by immunohistochemical staining with a primary antibody to all isotypes. Neither staining pattern nor percentage of stained cells could be used to differentiate the lesions studied. The statistically significant differences found between some groups are not likely to be of diagnostic value. p63 is not coexpressed with p53 expression or Ki-67 suggesting functional independence. When antibodies to the p63 isotypes become available, oral dysplasias and carcinomas should be reassessed.