Around 20% of patients with primary high-grade ovarian cancer do not respond to chemotherapy, but predictive biomarkers are lacking. The purpose of the current study is to establish and clinically validate an ex vivo 3D micro-tumour testing platform that predicts patient-specific response to standard of care chemotherapy. 104 ovarian cancer patients with malignant ascites were included in the study. Micro-tumours enriched from ascites were exposed to standard of care chemo- and targeted therapies, imaged using a high-content 3D screening platform. Morphological features were extracted for sensitivity profiling. A linear regression model was trained to predict the patient's CA125 decay rates, which were correlated to clinical outcomes (patient CA125 decay rate, change in tumour size, and progression-free survival). Isolated micro-tumours recapitulated ovarian cancer markers. A significant correlation (R = 0.77) between predicted and clinical CA125 rates was observed. Patients with predicted high ex vivo sensitivity to carboplatin/paclitaxel demonstrated significantly increased PFS and decreased tumour size. Complementary, patient-specific response profiles for second-line therapies were calculated and presented in integrated reports. In conclusion, an ex vivo 3D micro-tumour testing platform was established that predicted clinical response to neo-adjuvant chemotherapy in ovarian cancer patients and measured patient-specific responses to second-line therapies as a proof-of-concept. The platform enabled stratification of responders vs non-responders and has the potential to support informed treatment decisions after prospective validation. Results are generated within 2 weeks after sample collection, aligning with the clinical time frame for treatment decision-making.