Abstract The aim was to test the null-hypothesis that there is no difference in the cumulative survival rate of retained composite resin (CR) sealants and a high-viscosity glass-ionomer Atraumatic Restorative Treatment (ART) sealant in first permanent molars calculated according to the traditional and the modified retention assessment criteria over a period of 3 years. This cluster-randomized controlled clinical trial consisted of 123 schoolchildren, 6–7-years-old. At baseline, high-caries risk pits and fissures of fully erupted first permanent molars were treated with CR and ART sealants. Evaluations were performed after 0.5, 1, 2 and 3 years. Retention was scored for free-smooth surface and for each of three sections into which the occlusal surface had been divided. The modified criterion differed from the traditional in that it determined an occlusal sealant to be a failure when at least one section contained no visible sealant material. Data were analysed according to the PHREG model with frailty correction, Wald-test, ANOVA and t-test, using the Jackknife procedure. The cumulative survival rates for retained CR and ART sealants in free-smooth and occlusal surfaces for both criteria were not statistically significantly different over the 3 years. A higher percentage of retained CR sealants on occlusal surfaces was observed at longer evaluations. Cumulative survival rates were statistically significantly lower for the modified criterion in comparison to the traditional. The modified retention assessment criterion should be used in future sealant-retention studies.