As genomic analysis technology has advanced, it has become possible to sub-classify intrahepatic cholangiocarcinoma (ICC) at the histological or molecular level. However, there are no truly representative models of ICC subtypes for use in studying developmental differences, carcinogenesis, and personalized drug response. Here, we sought to develop ICC organoid models that reflect the phenotypic, molecular and genetic properties of the parent tumor, and performed integrative clustering of multiple genomic data—with subtype analysis—using these models. ICC organoids from 17 patients pathologically diagnosed with cholangiocarcinoma were prepared according to a previously established organoid culture protocol. ICC patients were subclassified into small-duct (SD) type and large-duct (LD) type according to histological characteristics and S100P, N-cadherin, and CD56 expression. ICC organoids were successfully established within one month and exhibited a morphology similar to that of their matching primary cancer. LD- and SD-type organoids exhibited histologic phenotypes and staining patterns characteristic of the corresponding ICC subtypes. ICC organoids showed high concordance of somatic mutations with primary tumors. Unsupervised principal component analysis clustering effectively separated each type of ICC. Differential gene expression revealed significant enrichment on KRAS, TGFβ and ERBB2 signaling pathways in LD-type compared with SD-type ICC (P < 0.05). Gene set enrichment analysis further demonstrated that the cholangiocarcinoma class 2 signature, defined by Andersen et al., was significantly enriched in the LD-type (enrichment score = 2.19, P < 0.001). A protein-protein interaction network analysis identified ZNF217 as a significant hub protein (odds ratio = 4.96, P = 0.0105). We successfully performed prospective modeling of histological subtype specification using patient-derived ICC organoids. Moreover, gene expression profiling of ICC organoids enabled identification of type-specific targetable pathways.