Abstract Objectives. To validate a risk classification model according to the Public Dental Service (PDS) guidelines with the actual 3-year caries outcome in terms of predictive values. Materials and methods. All 19-year-old patients registered at eight public dental clinics in Skåne, Sweden were invited to participate (n = 1699). The study group who completed the baseline examination consisted of 1295 subjects, representing 10% of the age group attending the public dental care in the region. A risk classification of each patient in four categories was made by the patient's regular team according to the guidelines. At the follow-up 3 years later, 75.8% were accessible. The final examinations included bitewing radiographs and the actual caries increment for 982 patients was calculated. Results. At baseline, 11.9% were assessed as being at 'high' or 'very high' risk. The dropouts had significantly more caries than the remaining patients (p < 0.05). The general disease activity was low, but those grouped into the two highest risk groups displayed significantly more new caries than those at lower risk (p < 0.05). With a cut-off value ΔDFS > 0 vs DFS = 0, the sensitivity was 81% and the specificity 56% for 'low' risk vs any risk. By combining the 'low' and 'some' risk categories, the PDS model generated an improved specificity (94%). Conclusions. The risk assessment scheme used by the Public Dental Service for young adults relied basically on past and current caries activity and 65-70% of the patients were correctly classified. The model was most effective to screen out patients with low caries risk.