Purpose: To study the microstructure of sealant penetration in the enamel of in vivo sealed approximal noncavitated incipient caries lesions with and without a preceding bonding step. Materials and Methods: A total of 26 proximal noncavitated caries lesions were sealed in vivo, using a sealant in 13 premolars with orthodontic indication of extraction. Each tooth was randomly assigned to mesial or distal surface application of a sealant in the lesion area and in surrounding sound enamel, with or without a bonding system. Four groups were analyzed: a nonbonding group in the lesion area (NBL); a nonbonding group in sound enamel (NBS); a bonding group in the lesion area (BL) and a bonding group in sound enamel (BS). The premolars were extracted after two weeks. All sealed areas were cut and demineralized with 37% hydrochloric acid for 24 h. For each group, the resin tags were observed and measured by scanning electron microscopy. Results: The lesion areas showed a very irregular resin network with twisted and curved tags in contrast to the sound enamel where a regular etching pattern was observed. The length of resin tags in microns for each group (mean ± SD) was: NBG-L = 4.19 ± 1.59; NBG-S = 5.49 ± 2.49; BG-L = 4.57 ± 1.99; and BG-S = 4.21 ± 1.87. The differences between the groups were not statistically significant (p=0.34). Conclusion: The use of a bonding system prior to the application of a pit and fissure sealant on both lesion and sound enamel areas does not increase the resin penetration length under non-contaminated conditions. Keywords: approximal initial caries, in vivo, prevention, sealant, SEM analysis