Standardised epidemiological caries assessments used in oral health surveys have been shown to be poor at predicting whether a tooth surface will be treated restoratively when a patient visits a dentist. However, it has been argued that oral health surveys may be more relevant in determining needs at the level of an individual or groups of individuals. The objective of this study was to determine the discriminatory power of visual caries assessments at two thresholds (D1 & D3) in adolescents of average age 12.1 years to predict experience of dentinal caries 3 years later or the experience of restorative treatment (not re-treatment) during the 3-year period. The data was derived from a prospective 3-year longitudinal study in which the dental care provided by 41 dentists for 403 adolescents was monitored. Dental caries experience was monitored by annual standardised assessments of caries undertaken by a single trained examiner. ROC analysis showed that caries assessed visually at the D1 threshold in 12-year-olds was a better predictor (P < 0.001) of experiencing some dentinal caries after 3 years (Az = 0.781) than was caries assessed visually at D3 threshold in 12-year-olds (Az = 0.670). Assessing caries visually at either the D1 or the D3 threshold had no discriminatory power for predicting whether an individual would experience some restorative treatment during the ensuing 3-year period (Az for D1 = 0.507; Az for D3 = 0.518).