A cluster randomized controlled oral cancer screening trial is on-going in the Trivandrum district, India, to evaluate the efficacy of screening in reducing oral cancer mortality. Subjects, aged 35 years and above, in 13 clusters in the Trivandrum district, India, were randomized to the intervention (screening) group (7 clusters, 78969 subjects) to receive three rounds of screening by oral visual inspection by trained health workers at 3-year intervals or to a control group (6 clusters, 74739 subjects). Two rounds of screening were completed between 1995 and 2002 during which 69896 (88.5%) subjects were screened at least once, and 59.7% of the 4408 screen-positive subjects were further investigated. In the intervention group, 344404 person-years were accrued and 329326 person-years were in the control group. In the intervention group, 149 incident oral cancer cases and 65 deaths from oral cancer were observed, and 106 incident cases and 62 deaths from oral cancer were observed in the control group. The programme sensitivity for detection of oral precancerous lesions and cancer was 81.5% and the programme specificity was 84.8%; the programme positive predictive value was 39.6%. In the intervention group 37.6% of the cases were in stages I-II, as opposed to 18.9% in the control group. The 3 year survival rate was 57.5% in the intervention and 38.8% in the control group (P<0.05). The age standardized oral cancer mortality rates were 21.2/100000 person-years in the intervention and 21.3/100000 in the control group. After completing two rounds of screening, oral cancer mortality rates were similar in both study groups.