Oral cavity cancer is usually the result of heavy smoking and longstanding alcohol consumption. Only around 10% of oral cancer patients neither smoke nor drink. In western countries, there is a rising trend in oral cancer in those who neither smoke nor drink alcohol. Human papilloma virus (HPV) has already been linked to oropharyngeal cancer, especially in those who neither drink nor smoke. This relationship between HPV and oral cancer has not yet been established, with previous results being controversial. We have hypothesised that the frequency of high risk HPV types, especially types 16 and 18, is increased in non smokers non drinkers (NSND) who develop oral cancer. This would therefore link HPV involvement in oral carcinogenesis in this population. Only patients with oral squamous cell carcinoma treated primarily with surgery in the Department of Maxillofacial Surgery, University Hospital Dubrava, Zagreb, from 1995 to 2010 were included in the study. Patients were divided into two groups. The first group consisted of smokers of more than 10 pack/year who were also daily drinkers. The second group consisted of NSND and included those of less than 10 pack/year active smokers. HPV status, age, gender, anatomic locations of the tumour, stage, second primary tumours, recurrence and survival were compared between the two groups. HPV was detected directly with PCR method and indirectly with p16INK4a immunohistochemistry. There were 77 non smokers non drinkers and 77 controls (smokers and drinkers, SD). Women were more frequent among NSND group (75%) versus more males among SD (90%), p<0,001. The age range was greater among NSND, with the youngest patient 24 years old and the oldest one 90 years old. More than half of the patients were younger than 45 or older than 70 years. The SD group were middle aged (45-69 years) in 79%. The tongue was the most frequent tumour site in oral cavity among NSND (42%) followed by the sublingual region (17%), upper gum (13%) and lower gum (12%). In the control group (smokers and drinkers), sublingual cancer was the most frequent site (39%), followed by tongue cancer (34%) and then retromolar cancer (17%). The NSND group had smaller cancers (T1 and T2) without regional metastasis (N0), although there was no statistical significance. No difference in recurrence or second primary tumour (SPT) was observed between the two patient groups. HPV detected by PCR was less frequent among the NSND group. Only 22% of NSND were positive for HPV, compared with 31% of SD. HPV was positive in 27% of all oral cancer patients which is similar to previous published results. Men were more frequently HPV positive (68%) than women. HPV positive patients tended to be younger than HPV negative patients (56 years vs 60 years) but without statistical significance. Regional metastasis (N) correlated significantly more with HPV positivity (39% vs 23%) although there was no difference in primary tumour (T) or stage. Similar rate of SPT was observed between the two groups. HPV positive patients had a greater incidence of recurrences and worse survival, but this was not statistically significant. If p16INK4a expression is considered as surrogate marker of HPV infection, we have confirmed the hypothesis of the study that high risk HPV groups are more frequently seen in NSND oral cancer patients than in SD patients with oral cancer. p16INK4a was positive in 27% of NSND and only 10% of SD (p=0,0064). However, PCR results did not support the above hypothesis and the overall correlation between p16INK4a and HPV was weak. All p16INK4a positive patients should be HPV positive if it is sensitive as a surrogate marker, but this was the case in only 7 out of 29 patients. 22 out of 29 patients were p16INK4a positive and HPV negative. Based on our findings and those of other studies, we cannot support the use of p16INK4a immunostaining as the only test for HPV in oral cancer. No statistically significant difference was observed between p16INK4a positive and negative patients according to age, sex, anatomic tumour locations, stage, SPT or HPV status. A higher rate of recurrences was associated with stronger p16INK4a expression, and this trend was also observed in survival curves. Stronger p16INK4a expression had worse survival outcomes, particularly among tongue cancers. Both HPV PCR and p16INK4a may be extrapolated for survival and prognostic purposes. When both are positive, survival outcomes are worse while best survival is associated with negative HPV PCR and p16INK4a. NSND that is HPV positive has 2.8 hazard ratio and strong p16INK4a expression carries 2.9 hazard ratio for the same population. According to this study, correlation between HPV and p16INK4a is very weak. In clinical practice, patients should be routinely tested for HPV using PCR and p16INK4a immunohistochemistry in order to give us valuable information regarding their prognosis.