The aim of this study was to evaluate how cavity linings with different elastic modulus can influence the marginal adaptation (MA) of Class II composite restorations before and after thermo-mechanical loading. Forty Class II cavities with margins extending 1 mm below the cement-enamel junction were prepared in extracted human third molars. In each group except the control group, a lining material of 1-mm thickness was applied to the bottom of the cavity and polymerized before placing the resin composite Herculite XRV Ultra (group A: control; group B: Premise Flowable lining; group C: Herculite XRV Ultra lining; and group D: Optibond FL lining). MA was evaluated (with a scanning electron microscope) before and after loading (200,000 loading cycles). Statistical analysis was done using the Shapiro-Wilks test, the analysis of variance test, and Duncan post hoc test at p<0.05. Before loading, the percentages of continuous margins in dentin were superior (p<0.05) for groups C and D (71.1% and 87.2%, respectively) compared to groups A and B (55.7% and 48.3%, respectively). After loading, group D (79.8%) was statistically superior in dentin compared to all of the other groups (43.6%, 35.9%, and 54.4%, respectively). In occlusal enamel, no significant difference was found between groups. The percentage of enamel fractures and the percentage of noncontinuous margins in proximal enamel were high, with no significant difference between liners. It can be concluded that for the materials used in this study, a 1-mm-thick lining with an extremely low elastic modulus (2-3 GPa) could redistribute shrinkage stress. The use of a flowable composite did not significantly improve MA.