In spite of the latest advancements in understanding cancer development and progression, drugs successful in preclinical testing often fail upon reaching phase III clinical trials. A reason for this is the use of inappropriate preclinical models that do not preserve tumor heterogeneity. Although used for decades, cell cultures derived from patients substantially deviate from their original biopsy upon culturing; moreover, they cannot predict the response of an organism as a whole. Patient-derived xenograft (PDX) models are emerging as powerful tools since they have a predictive therapeutic value and preserve the heterogeneity of the original tumors. PDX are established by implanting freshly isolated tumors from patients into immunocompromised mice, allowing for the progressive growth and amplification of cancer tissue for in vivo testing. Here, we describe the detailed methods we developed to establish PDX from both surgically removed endometrial cancer fragments (endometrial cancer) and fine-needle aspiration biopsies (pancreatic cancer).