Can a functional in vitro model, containing the main cellular components of the uterine wall, be generated from cells derived from patient tissues? We present a three-dimensional (3D) physiologically relevant, organ-on-a-chip model of the uterine wall containing primary endometrial and myometrial cellular participants, generated from human uterine tissue. As a highly dynamic reproductive organ, the human uterus plays fundamental physiological roles in menstruation and childbirth. The endometrial-myometrial junction (EMJ) defines the interface between the inner mucosal layer (endometrium) and outer smooth muscle zone (myometrium) that comprises the uterine wall. The EMJ is implicit in several uterine pathologies of unknown aetiology, including adenomyosis and abnormally invasive placenta; however, despite this, no patient-derived in vitro models of the uterine wall containing all EMJ participants currently exist. We employed microfluidic technology to characterize multiple miniaturized models of the uterine wall. Protocols were tested that included variations in the seeding order of endometrial and myometrial fractions, and the addition of a low viscosity extracellular matrix to influence cell behaviour. Ultimately, functional hormone responses of patient-derived uterine wall models were assessed. Endometrial (n = 9) and myometrial biopsies (n = 4) were enzymatically dissociated to create epithelial, stromal and myometrial cellular fractions. Cell suspensions were seeded into non-adhesive poly(dimethylsiloxane) microfluidic devices containing 5 × 5 microwell arrays. The fate of individual cell types was monitored in real-time using fluorescent tracers, and cell phenotype was characterized by immunocytochemistry. Model functionality was assessed by measuring Ca2+ responses to agonist stimulation, and both insulin-like growth factor binding protein 1 (IGFBP-1) and osteopontin secretion in response to hormone stimulation. When subjected to microfluidic culture in isolation, endometrial stromal cells and smooth muscle myocytes formed compact spheroids, whilst epithelial cells produced diffuse aggregates. Tri-cultures were established by sequential seeding of individual or combined cell fractions at various ratios. Regardless of the protocol, epithelial cells localized to the outer periphery of tri-culture spheroids, which varied in morphology across the protocols. Incorporation of 5% [v/v] Matrigel® improved the reproducibility of 3D aggregates which exhibited robust self-assembly of a stromal/smooth muscle core encased in epithelium. Exposure of tri-cultures to oestradiol, medroxyprogesterone acetate and cyclic adenosine monophosphate (cAMP) increased secretion of IGFBP-1, which indicates stromal decidualization, and enhanced epithelial cell osteopontin secretion. Stimulation with endothelin-1 induced Ca2+ signalling in myocytes. Endometrial and myometrial tissue was collected from relatively few donors. Myometrial tissue was collected from pregnant donors, which may have influenced the myocyte phenotype. Furthermore, endometrial tissue sampling was from women not having a hysterectomy, thus may not include the deeper basalis region, which may limit the physiological mimicry of the final models. Our novel approach to modelling the uterine wall in 3D captures all of the main cell types in a medium-throughput system, enabling the screening of hundreds of cultures in parallel from a single biopsy. This system shows great promise for examining the cellular interplay between physiological cues and EMJ pathologies, such as the impact of uterine peristalsis and cyclical hormones on the pathogenesis of adenomyosis. C.B. was supported by an Organ-on-a-Chip Technologies Network Pump Priming Project grant. C.J.H. was supported by a Wellbeing of Women project grant (RG2137), SRI/Bayer and Wellcome Trust IFFS3. D.K.H. was supported by a Wellbeing of Women project grant (RG2137) and MRC clinical research training fellowship (MR/V007238/1). M.Z. is Director and Co-Founder of ScreenIn3D Limited. The other authors declare no conflict of interest. N/A.