OBJECTIVE: This in vitro study aimed to evaluate occlusal caries extension in relation to visual and radiographic diagnostic criteria and their clinical value to indicate operative or preventive dental care. METHODS: A total of 196 third molars with clinically sound occlusal fissures or noncavitated lesions were collected. Before microcomputed tomography (µCT) investigation, each tooth was examined visually and radiographically. Kühnisch's µCT-based caries-extension index (CE index) was used to determine the caries depth on a numeric scale (0 = sound; 0.01-0.99 = enamel caries; 1.0-1.99 = dentin caries). Sensitivities (SEs), specificities (SPs), and area under the receiver operating characteristic curve (Az value) were also calculated. RESULTS: Based on µCT data, the following mean CE index values and standard deviations (SDs) were documented according to the visual criteria: sound = 0.6 (0.4); first visible signs = 0.9 (0.4); established lesions = 1.3 (0.3); microcavities = 1.4 (0.2); dentin exposure = 1.5 (0.2); and large cavities = 1.5 (0.3). The radiographic categories according to Marthaler (enamel caries [D0-2], caries in the outer half of dentin [D3], and caries in the inner half of dentin [D4]) were related to CE index values of 0.9 (0.4), 1.4 (0.2) and 1.6 (0.4), respectively. Caries detected visually or radiographically showed an SE of 84% and an SP of 85% (Az = 0.85). When both methods were used to predict dentin involvement simultaneously, SE = 27%, SP = 100%, and Az = 0.63; this combined visual and radiographic approach was associated with a perfect specificity and no false-negative decisions. The proportion of false-positive diagnoses was moderately high, and lesion extension in these cases was mainly limited to the outer 20% of the dentin. CONCLUSIONS: Our results might be useful for differentiating between preventive and operative dental care for pits and fissures.