To evaluate methods of preventing young children from experimenting with tobacco and to determine cost effectiveness, students (n = 1005) in 31 primary schools, from randomly selected higher gradelevels were recruited into a partially randomized, single blinded controlled trial in which seven groups of schools were randomly assigned to a combination of teaching, leaflet, and drama, in order to modify students’ knowledge, attitudes, beliefs and behaviour (KAB). The eighth group (n = 346) with ten schools, distantly separated from the former, was assigned to be the control, but was dropped from comparison analysis for lack of randomness at baseline. The mean, standard deviation and median age of the intervention groups was 9.94 years (0.81), 10.0 years, (n = 669) at baseline; and 10.62 years (0.66), 11.0 years, (n = 397), at 12 months follow-up. In all, 6.6% had ever used tobacco at least once at a median age of seven years. Teaching health education at school when combined with other methods was significantly better at improving KAB. In 2003, after a year post-intervention, the occurrence of experimentation smoking in the last 30 days, dropped from 9.2% to 1.2% (p = 0.00), equivalent to 87% (95% CI 78, 93) reduction in the group exposed to health education compared to none in the leafletonly group and Numbers Needed to Treat (NNT) = 12.5. Due to its cost-effectiveness (comparable to child immunizations) at BDS$1.89 to 2.89 or US$1 to 1.5 per child contacted and BDS$100 to 140 (US$50 to 70) capital investment in other resources per school, the experience could be utilized routinely in elementary schools.