Evidence on the caries-preventive effect of chlorhexidine (CHX) among high-risk children is inconclusive, possibly because obscured by fluoride exposure. We investigated the effect of CHX among initially 3-year-old subjects whose baseline d3ft was = 0 and whose only regular fluoride exposure came from toothpaste. The subjects were assigned to three groups: high-risk test (HRT, n = 70), high-risk control (HRC, n = 71), and low-risk control (LRC, n = 70). Risk classification was based on salivary mutans streptococcal levels (MS, ≧1.0 × 105 cfu/ml). Basic measures (oral hygiene, dietary counselling every 4 months) were given to all groups. HRT also underwent CHX gel applications for 3 consecutive days at 3-month intervals for 15 months. Eighteen months after baseline d3ft increments and proportions of children with d3ft increment ≧1 (%d3ft increment ≧1) among all groups were assessed. Anti-MS effect on high-risk children and caries-preventive effect on all children were statistically analysed by residual change analysis (MS), non-parametric tests and logistic regression analysis (caries). No differences were found between the groups in basic programme compliance. CHX significantly reduced MS levels. %d3ft increment ≧1 and mean d3ft increments were 34.3%, 0.56 (HRT), 32.4%, 0.54 (HRC) and 11.4%, 0.11 (LRC), with HRT/HRC values statistically significantly higher than LRC values and no significant difference between HRT and HRC. HRT children were not less likely to show new lesions than HRC children (OR = 1.09; 95% confidence interval 0.54–2.19), while high-risk children were 4 times more likely to show new lesions than low-risk children (OR = 3.71; 95% confidence interval 1.53–9.03). CHX gel applications showed moderate anti-MS effect but negligible caries-preventive effect.