In this study, the effect of three different feedstocks-namely, straight-run atmospheric residue (AR), demetallized AR (DM-AR), and demetallized/desulfurized AR (DMDS-AR)-on the coke formation on an industrial catalyst system consisting of a hydrodemetallization (HDM) catalyst (A), a hydrodemetallization/hydrodesulfurization (HDM/HDS) catalyst (13), and a hydrodesulfurization/hydrodenitrogenation (HDS/HDN) catalyst (C) were examined. The used catalysts, aged under the same operating conditions (catalyst A with atmospheric residue (AR), B with AR or demetallized atmospheric residue (DM-AR), C with AR or demetallized and desulfurized atmospheric residue (DMDS-AR)), were characterized by combining physical/chemical analyses with temperature-programmed oxidation/mass spectroscopy (TPO/MS) and solid-state C-13 nuclear magnetic resonance (NMR). Special emphasis was made on finding a relationship between feed quality and the amount and nature of coke. The results revealed that hydrotreated feeds generate significantly more coke than straight-run AR. The higher propensity in coke formation using a demetallized feedstock is due to the bulky aromatic structure of the asphaltenes in DMAR, along with a high H/C ratio of the feed (phase separation-precipitation-coke formation). In the case of DMDS-AR, coke generation is attributed to the nature of catalyst C (namely, high acidity and small pores), rather than to the quality of the feed. Concerning the coke properties, it was found that the coke generated from hydrotreated feeds is less aromatic than that from AR. Regarding other coke features, such as degree of condensation or degree of alkyl substitution, no obvious interplay between these structural parameters and the nature of feed or catalyst could be found.