Prostate cancer (PCa) patient-derived xenografts (PDXs) are commonly propagated by serial transplantation of "pieces" of tumour in mice, but the cellular composition of pieces is not standardised. Herein, we optimised a microwell platform, the Microwell-mesh, to aggregate precise numbers of cells into arrays of microtissues, and then implanted the Microwell-mesh into NOD-scid IL2γ-/- (NSG) mice to study microtissue growth. First, mesh pore size was optimised using microtissues assembled from bone marrow-derived stromal cells, with mesh opening dimensions of 100×100 μm achieving superior microtissue vascularisation relative to mesh with 36×36 μm mesh openings. The optimised Microwell-mesh was used to assemble and implant PCa cell microtissue arrays (hereafter microtissues formed from cancer cells are referred to as microtumours) into mice. PCa cells were enriched from three different PDX lines, LuCaP35, LuCaP141, and BM18. 3D microtumours showed greater in vitro viability than 2D cultures, but neither proliferated. Microtumours were successfully established in mice 81% (57 of 70), 67% (4 of 6), 76% (19 of 25) for LuCaP35, LuCaP141, and BM18 PCa cells, respectively. Microtumour growth was tracked using live animal imaging for size or bioluminescence signal. If augmented with further imaging advances and cell bar coding, this microtumour model could enable greater resolution of PCa PDX drug response, and lead to the more efficient use of animals. The concept of microtissue assembly in the Microwell-mesh, and implantation in vivo may also have utility in implantation of islets, hair follicles or other organ-specific cells that self-assemble into 3D structures, providing an important bridge between in vitro assembly of mini-organs and in vivo implantation.