Experience gained from previous jet noise studies with the unstructured large-eddy simulation flow solver “Charles” is summarized and put to practice for the predictions of supersonic jets issued from a converging–diverging round nozzle. In this work, the nozzle geometry is explicitly included in the computational domain using an unstructured body-fitted mesh. Two different mesh topologies are investigated, with emphasis on grid isotropy in the acoustic source-containing region, either directly or through the use of adaptive refinement, with grid size ranging from 42 to 55×106 control volumes. Three different operating conditions are considered: isothermal ideally expanded (fully expanded jet Mach number of Mj=1.5, temperature of Tj/T∞=1, and Reynolds number of Rej=300,000), heated ideally expanded (Mj=1.5, Tj/T∞=1.74, and Rej=155,000), and heated overexpanded (Mj=1.35, Tj/T∞=1.85, and Rej=130,000). Blind comparisons with the available experimental measurements carried out at the United Technologies Research Center for the same nozzle and operating conditions are presented. The results show good agreement for both the flow and sound fields. In particular, the spectra shape and levels are accurately captured in the simulations for both near-field and far-field noise. In these studies, sound radiation from the jet is computed using an efficient permeable formulation of the Ffowcs Williams–Hawkings equation in the frequency domain. Its parallel implementation is reviewed and parametric studies of the far-field noise predictions are presented. As an additional step toward best practices for jet aeroacoustics with unstructured large-eddy simulations, guidelines and suggestions for the mesh design, numerical setup, and acoustic postprocessing steps are discussed.