Our drug discovery model has identified two novel STAT3 SH2 domain inhibitors 323–1 and 323–2 (delavatine A stereoisomers) in a series of experiments. In silico computational modeling, drug affinity responsive target stability (DARTS), and fluorescence polarization (FP) assays altogether determined that 323–1 and 323–2 directly target the STAT3 SH2 domain and inhibited both phosphorylated and non-phosphorylated STAT3 dimerization. Computational docking predicted that compound 323s bind to three subpockets of the STAT3 SH2 domain. The 323s inhibition of STAT3 dimerization was more potent than the commercial STAT3 SH2 domain inhibitor S3I-201 in the co-immunoprecipitation assay, correlating with computational docking data. The fluorescence polarization assay further confirmed that the compound 323s target the STAT3 SH2 domain by competitively abrogating the interaction between STAT3 and the SH2-binding peptide GpYLPQTV. Compared with S3I-201, the 323 compounds exhibited stronger inhibition of STAT3 and reduced the level of IL-6-stimulated phosphorylation of STAT3 (Tyr705) in LNCaP cells over the phosphorylation of STAT1 (Tyr701) induced by IFN-ɣ in PC3 cells or the phosphorylation of STAT1 (Ser727) in DU145 cells. Both compounds downregulated STAT3 target genes MCL1 and cyclin D1. Thus, the two compounds are promising lead compounds for the treatment of cancers with hyper-activated STAT3.