This in vitro study compared the proximal marginal adaptation of direct com
posite restorations with composite and ceramic inlays inserted with differe
nt resin cements. Standardized MOD Class II inlay cavities with one proxima
l box extending below and the other above the CEJ were cut in 48 extracted
human molars and randomly assigned to six groups (n=8). Incrementally layer
ed direct composite restorations (P60), composite inlays (P60) and ceramic
inlays (Empress; Cerec Vitablocs Mark II) were placed in the cavities. Thre
e different resin cements (RelyX ARC; Variolink II high viscosity; Panavia
21) were used for luting the composite inlays. All ceramic inlays were ceme
nted with RelyXARC. After finishing and polishing, the teeth were stored fo
r 24 hours in Ringer solution at 37 degreesC before they were subjected to
thermal and mechanical loading (5/55 degreesC, 2000x; 50 N vertical load, 5
0000x). Margins were evaluated on epoxy replicas using a scanning electron
microscope at X200 magnification. Statistical analysis was performed with n
on-parametric test methods (alpha =0.05). The adhesive interfaces to enamel
exhibited high percentages of perfect margins for all groups (91.8% to 96%
) and a maximum of 5.2% marginal gap formation. Dentin-limited cavity segme
nts demonstrated more marginal openings and less perfect margins than ename
l-bound areas; however, this was only statistically significant for direct
composite restorations and composite inlays inserted with Variolink II and
Panavia 21. RelyXARC showed a significantly better adaptation to P60 inlays
compared with the leucite-reinforced Empress ceramic but not the Vitablocs
Mark II ceramic.