An interesting question for philosophy of science is how the “personal” gets constituted, scientifically as well as socially, through new technologies and practices in personalized medicine. A novel approach to better account for patient variation is to develop so-called tumor organoids based on tumor samples from individual cancer patients. Given their ability to recapitulate tumor heterogeneity, patient-derived models have been highlighted as breaking way for a “science of the individual” or a “one patient paradigm” in medicine. But to what extent is it possible – and desirable – for in vitro models to become “substitutes” for patients or patient types? To explore such questions, we combine philosophical and ethnographic analysis of laboratory research and clinical research practice. We analyze how epistemic uncertainties about the evidential status of organoids relate to ontological uncertainties about the nature of cancer itself, and document challenges of determining what level of variation is scientifically and clinically meaningful in personalized medicine. Moreover, we show how epistemic and ethical implications intersect when tumor organoids are attempted used for patient-specific drug screening. In this context, researchers and clinicians become stretched between the hopes of patients and epistemic uncertainty.