Ovarian cancer (OvCa) is an insidious disease: due to nonspecific symptoms, these tumors are usually diagnosed at advanced stage with correspondingly devastating consequences for treatment outcome and patient survival. The broad heterogeneity of OvCa manifests itself in the complex composition of the tumor microenvironment. Given the frequent development of therapeutic resistance, there is a strong need for model systems accurately representing OvCa heterogeneity, while enabling parallel drug testing and prediction of appropriate treatment responses in individual patients. Here, we demonstrate the efficient isolation of highly viable OvCa patient-derived microtumors (OvCa PDM). Importantly, our data demonstrate histopathological comparability of OvCa PDM with corresponding patient tumor tissue. Reverse phase protein array (RPPA)-based analyses of >110 total and phospho-proteins enabled the identification of sensitivities to standard, platinum-based as well as experimental, selumetinib-based therapy, and thereby the prediction of treatment-responder. Parallelized drug testing in OvCa PDM allowed functional validation of RPPA data and detection of on- and off-target treatment effects. Strikingly, clinical follow-up of corresponding patients confirmed significantly increased metastasis-free survival of identified carboplatin-responder. Furthermore, flow cytometry-based characterization of autologous TIL populations confirmed the presence of tumor-specific, cytotoxic TILs with stem-like CD39-PD1+ and terminally differentiated CD39+PD1+ phenotypes. Interestingly, our results showed a significant correlation between the presence of CD8+CD39+ Tils and lymph node metastasis in the associated patients. Finally, combining OvCa PDM and autologous TILs for efficacy testing of immune checkpoint inhibitors demonstrated the potential for patient-specific enhancement of cytotoxic TIL activity by this therapeutic approach.