By means of a number of computationally more advanced methods the search for an acceptable overall calculated gas-phase geometry of Roesky's ketone (5-oxo-1,3,2,4-dithiadiazole) is continued. The results of CCSD, QCISD and MP4(SDQ) calculations are compared with the results of different CASSCF and DFT calculations. The results obtained with the wave-function-based methods are better than those generated by a large number of different DFT functionals, especially for the description of the carbon-sulfur bond. However, even at the CCSD and QCISD levels of theory no convergence is achieved: upon increasing the level of theory from CCSD to CCSD(T) the quality of the description actually becomes worse. (c) 2005 Elsevier B.V. All rights reserved.