A cyclic peptide was isolated from the deep-sea marine sponge Discodermia japonica, and its NMR spectroscopic data were identical to those reported for cyclolithistide A, a known antifungal depsipeptide. However, the interresidue HMBC correlations suggested that the amino acid sequence was different from that of the original structure. Moreover, chiral-phase GC-MS, combined with Marfey's analysis, indicated that the absolute configurations of three amino acids were also antipodal. Here, we propose the revised structure of cyclolithistide A and address the configuration of the previously unassigned 4-amino-3,5-dihydroxyhexanoic acid (Adha) moiety.