This study evaluated the effect of contamination media on the adhesion of resin cement to resin composites. Specimens of direct (DRC) (Quadrant Photo Posterior) and indirect resin composite (IRC) (Gradia) (N = 300, n = 15 per group) were prepared. Except the control group (C), the specimens of DRC and IRC were contaminated with one of the following media: (a) saliva (S), (b) silicon (SI), (c) dental stone (D), and (d) isolation medium (I). While one half was only rinsed with water, the other half was silica coated (30 μm SiO2, Siljet). All specimens were silanized (Monobond Plus) and coated with adhesive resin (Heliobond). Resin cement (Variolink II) was bonded to the substrates and photo-polymerized for 40 s. After thermocycling (x5.000, 5–55 °C), composite–cement interface was loaded under shear in a Universal Testing Machine (1 mm/min). Data (MPa) were analyzed using Univariate analysis, Tukey’s and Dunnett-T3 tests. Both contamination media (p = 0.000) and surface conditioning (p = 0.005) significantly affected adhesion to DRC and IRC. No significant difference was found between the DRC and IRC (without: p = 0.098; with: p = 0.084). Significantly lower results were obtained after SI (DRC: 0.66 ± 0.6; IRC: 0.8 ± 1.3) followed by I contamination (DRC: 2.1 ± 2.6; IRC: 0.8 ± 1.3) (p < 0.05). Regardless of contamination medium, surface conditioning significantly increased the results for both DRC (15.1 ± 6.1–23.6 ± 3.7) and IRC (20.3 ± 5.4–25.1 ± 3.6) (p < 0.05). Weibull distribution increased after surface conditioning for both DRC (without: 1.33–3.27; with: 2.55–9.34) and IRC (without: 1.07–3.75; with: 3.7–7.73). Predominantly adhesive (132 out of 150) failures were observed when surfaces were not conditioned.