<h3>Introduction:</h3> Cutaneous T-cell lymphomas (CTCLs) are a clinically heterogeneous array of clonal neoplasms of skin-homing CD4+ T-helper lymphocytes. Mycosis fungoides (MF) is the most common subtype of CTCL. Its clinical course varies significantly. Median survival times range from over fifteen years in its mildest form to under two in its most severe. Elucidation of MF pathogenesis and drug development are hampered by the dearth of CTCL laboratory models. Models examined in the literature are largely derived from leukemic CTCL, many of which have been shown to be confounded by HTLV-1 virus positivity. Additionally, recent models demonstrate phenotypic divergence even in the context of stable TCR clonality across passages. Here we molecularly characterize patient-derived xenografts (PDX) from six MF patients, compare their features to known CTCL characteristics and examine their stability over passages. <h3>Materials and methods:</h3> We established and analyzed ten PDX lines from six MF patients. Patient tumor cells were obtained from peripheral blood or lymph node fine needle aspirate and engrafted into NSG mice (P0). Subsequent passages (P1/2) were made by engrafting cells from P0/1 spleens in naive NSG mice. Peripheral blood mononuclear cells (PBMCs) from a subset of patients were collected for comparison to PDXs. Whole exome sequencing (WES) data was used for mutational and copy number analysis. Three PDX lines were also transcriptionally characterized using RNA sequencing. <h3>Results:</h3> WES analysis demonstrates the presence of canonical CTCL driver events in all samples. In all PDXs with patient PBMCs available for comparison, we find that subsets of these driver events were maintained through grafting. Strikingly, PDXs established from tumor cells collected at later points in clinical disease progression reflect tumor evolution. We identify genomic features that may show associations with successful engraftments. Through RNA-seq analysis, we observe that PDX transcriptional phenotypes harbored similarities to primary patient samples in the literature. We also demonstrate the transcriptional stability of one PDX into P1. <h3>Conclusion:</h3> Overall, we demonstrate that we have established PDXs that faithfully retain genomic and transcriptional features of CTCL across passages. Our results have implications not only for use of the models described here, but also for development of future xenografts by similar methods. We predict that employing molecularly stable CTCL PDXs in preclinical experiments could have significant benefits for the elucidation of CTCL pathophysiology and the development of novel CTCL treatments.