Objective To examine the relationship between the ICCMS™ (International Caries Classification and Managing System) features of root caries lesions and the underlying depth of the lesion towards the pulp. In order to control for bias, the study followed the tailored document for risk of bias assessment (RoB-tool) recently published. A pilot study showed that the outline of the pulp was much clearer on horizontal compared to vertical sections through the lesions ( p  = 0.03) and that the histological stereomicroscopical (SM) assessed lesion depth towards the pulp was not influenced by the cutting direction ( p  = 0.155). Material and methods A sample of extracted permanent molar teeth ( n  = 100) were classified independently by two of the authors according to ICCMS™ as no sign of root caries lesion 0 = sound; 1 = initial lesion (non-cavitated); 2 = moderate lesion (cavity depth ≤ 2 mm) and 3 = extensive lesion (cavity depth > 2 mm). After horizontal sectioning (HS) through the lesion, the depth of the underlying lesion was SM assessed independently by two of the authors as 0 = no lesion; 1 = lesion in outer 1/3; 2 = middle 1/3; and 3 = inner 1/3 of the dentine towards the pulp. Results Intra- and inter-reproducibility (weighted kappa values ≥ 0.83); the accuracy (Spearman’s rho-values) = 0.94 and 0.95; and specificity/sensitivities/AUC values (three different thresholds) were ≥ 0.91, ≥ 0.93, and ≥ 0.96, respectively. Conclusion Under the umbrella of the RoB-tool, the validity in terms of the reproducibility and accuracy of the ICCMS™ root caries scoring system was high. Clinical relevance By means of the ICCMS™ root caries scoring system, the underlying lesion depth can be estimated, which must be considered when managing the lesion.