The present study evaluated the influence of various low-shrinkage restorative systems in class II direct composite restorations following simulated occlusal loading. Forty MOD class II cavities were prepared on freshly extracted human lower third molars with proximal margins located mesially 1.0 mm coronal to and distally 1 mm apical to the cementoenamel junction. The samples were randomly distributed into five experimental groups corresponding to the following restorative systems: a conventional resin composite (Tetric) as active control group, a low-shrinkage composite (Extra Low Shrinkage [ELS]) alone or combined with its corresponding flowable version (ELSflow) used as a 1- to 1.5-mm liner, a bulk-filling flowable composite (Surefil SDR) covered by a 1-mm layer of restorative composite (Ceram-X), and a restorative bulk-filling composite (SonicFill). All specimens were submitted to 1,000,000 cycles with a 100N eccentric load into saline. Tooth restoration margins were analyzed semiquantitatively by scanning electron microscopy before and after loading. The percentage of perfect adaptation to enamel varied from 94.15% (SonicFill) to 100% (ELS) before loading and from 69.22% (SonicFill) to 93.61% (ELS and ELSflow) after loading. Continuous adaptation to cervical dentin varied from 22.9% (Tetric) to 79.48% (SDR/Ceram-X) before loading and from 18.66% (Tetric) to 56.84% (SDR/Ceram-X) after loading. SDR/CeramX and SonicFill showed the best cervical dentin adaptation.