Abstract Objectives Previous studies have shown good correlation between polycystic ovarian morphology (PCOM) and serum anti-Müllerian hormone (AMH) levels. We evaluated the utility of AMH as a surrogate for PCOM as a part of the polycystic ovary syndrome (PCOS) diagnosis by describing how the use of different AMH cut-off values would change the prevalence of PCOS. Methods A general population-based birth cohort study. Anti-Müllerian hormone concentrations were measured from serum samples taken at age 31 years (n = 2917) using the electrochemiluminescence immunoassay (Elecsys). Anti-Müllerian hormone data were combined with data on oligo/amenorrhoea and hyperandrogenism to identify women with PCOS. Results The addition of AMH as a surrogate marker for PCOM increased the number of women fulfilling at least two PCOS features in accordance with the Rotterdam criteria. The prevalence of PCOS was 5.9% when using the AMH cut-off based on the 97.5% quartile (10.35 ng/mL) and 13.6% when using the recently proposed cut-off of 3.2 ng/mL. When using the latter cut-off value, the distribution of PCOS phenotypes A, B, C, and D was 23.9%, 4.7%, 36.6%, and 34.8%, respectively. Compared with the controls, all PCOS groups with different AMH concentration cut-offs showed significantly elevated testosterone (T), free androgen index (FAI), luteinizing hormone (LH), LH/follicle-stimulating hormone (FSH) ratio, body mass index (BMI), waist circumference, and homoeostatic model assessment of insulin resistance (HOMA-IR) values, as well as significantly decreased sex hormone-binding globulin (SHBG) values. Conclusions Anti-Müllerian hormone could be useful surrogate for PCOM in large data sets, where transvaginal ultrasound is not feasible, to aid the capturing of women with typical PCOS characteristics. Anti-Müllerian hormone measurement from archived samples enables retrospective PCOS diagnosis when combined with oligo/amenorrhoea or hyperandrogenism.