Polycystic ovary syndrome (PCOS) is a multifaceted condition with diagnostic challenges and clinical heterogeneity across populations. Research priorities include enhanced accuracy in defining cut-offs for diagnostic features. Here, we aim to describe participant clinical features and data harmonization in the international PCOS Phenotype in Unselected Populations (P-PUP) study.We searched EMBASE and Medline (Ovid) from 1990 to October 2, 2020, in population-based, medically unbiased study cohorts. Included studies had ≥ 300 participants, directly assessed PCOS-related features, and provided Individual Participant Data (IPD). Risk of bias was assessed using the AXIS tool. Data integrity was ensured via cross-referencing, identifying outliers/implausible data, and variable harmonization. Reporting follows PRISMA-IPD guidelines, summarizing findings with frequencies and proportions.The study included 9979 reproductive-age women from 12 studies across eight countries (China, Iran, Italy, Nigeria, Russia, South Korea, Turkey, and the USA), representing 11 ethnicities. Ovulatory dysfunction was variably recorded, from mean menstrual cycle length, minimum or maximum cycle length, number of cycles per year, or urinary progesterone measurements. Clinical hyperandrogenism was assessed via modified Ferriman-Gallwey (mFG) scores, with a few also including acne and alopecia. Biochemical hyperandrogenism thresholds varied (95th, 97.5th, or 98th percentile of healthy controls). Polycystic ovary morphology was assessed via transvaginal, transabdominal, or transrectal approaches. Harmonization adhered to International PCOS Guidelines for ovulatory dysfunction, ethnicity-specific cut-offs for hirsutism (via k-means clustering), and 95th percentile thresholds for biochemical hyperandrogenism. PCOS prevalence ranged from 3.3 to 19.8% in the original studies and was 11.0% overall after harmonization.The P-PUP study offers an unprecedented, ethnically diverse, medically unbiased population-based cohort, an extraordinarily valuable tool to enhance knowledge and research in PCOS. However, variability in data collection methods and definitions of PCOS diagnostic features across studies limited the ability to fully integrate data for analysis. Despite these limitations, we optimized harmonization in this IPD, and the findings provided valuable insights into the challenges of data harmonization and established a foundation for future collaborative research. Future research should focus on standardizing data collection, establishing normative cut-offs based on true natural groupings, and linking diagnostic clusters to outcomes in diverse populations.CRD42021267847.© 2025. The Author(s).