This study compared the surface roughness of selected tooth coloured restorative materials that were polished according to manufacturers’ instructions and Sof-Lex. It also assessed the surface roughness of polished materials after thermocycling. Filtek Z350XT, Beautifil-Bulk Restorative and Cention N, were used in this study. A stainless steel mould (10mm diameter x 2mm height) was used to fabricate 75 cylindrical specimens: 15 Filtek Z350XT (FZ), 30 Beautifil-Bulk Restorative (BB) and 30 Cention N (CN). All 15 FZ specimens were polished with Sof-Lex. Fifteen BB and CN specimens were polished according to manufacturers’ instructions. The remaining fifteen BB and CN specimens were polished using Sof-Lex. All the specimens were subjected to thermocycling (1000 cycles). Surface roughness was assessed quantitatively with profilometry after specimen preparation (Mylar stage), polishing and thermocycling. Data were analysed using SPSS version 25.0 at α=0.05. When polished according to manufacturers’ instructions, BB had the lowest mean surface roughness (Ra) values (0.13±0.01μm) followed sequentially by CN (0.14±0.03μm) and FZ (0.15±0.02μm). The differences were not statistically significant. When polished with Sof-Lex, BB exhibited the smoothest surface (0.116±0.03μm) followed sequentially by and FZ (0.150±0.02μm) and CN (0.157±0.02μm). Thermocycling caused an increase in the Ra. The differences were statistically significant. All materials tested had Ra values below the threshold value of 0.2 µm at Mylar stage and after polishing with their recommended polishing system and Sof-Lex. Thermocycling produced rougher surfaces that did not exceed the threshold Ra value. Polishability was material dependent.