Abstract STUDY QUESTION Is it possible to identify a reliable marker of successful sperm retrieval (+SR) in men with idiopathic non-obstructive azoospermia (iNOA) undergoing microdissection testicular sperm extraction (mTESE)? SUMMARY ANSWER A higher likelihood of +SR during mTESE is observed in men with iNOA and lower preoperative serum anti-Müllerian hormone (AMH) levels, with good predictive accuracy achieved using an AMH threshold of <4 ng/ml. WHAT IS KNOWN ALREADY AMH has been previously linked to +SR in men with iNOA undergoing mTESE prior to ART. STUDY DESIGN, SIZE, DURATION A multi-centre cross-sectional study was carried out with a cohort of 117 men with iNOA undergoing mTESE at three tertiary-referral centres. PARTICIPANTS/MATERIALS, SETTING, METHODS Data from 117 consecutive white-European men with iNOA presenting for primary couple’s infertility associated with a pure male factor at three centres were analysed. Descriptive statistics was applied to compare patients with negative (−SR) versus +SR at mTESE. Multivariate logistic regression models were fitted to predict +SR at mTESE, after adjusting for possible confounders. Diagnostic accuracy of the factors associated with +SR was assessed. Decision curve analyses were used to display the clinical benefit. MAIN RESULTS AND THE ROLE OF CHANCE Overall, 60 (51.3%) men had an −SR and 57 (48.7%) had a +SR at mTESE. Patients with +SR had lower levels of baseline AMH (P = 0.005) and higher levels of estradiol (E2) (P = 0.01). At multivariate logistic regression analysis, lower levels of AMH (odds ratio: 0.79; 95% CI: 0.64–0.93, P = 0.03) were associated with +SR at mTESE, after adjusting for possible confounders (e.g. age, mean testicular volume, FSH, and E2). A threshold of AMH <4 ng/ml achieved the highest accuracy for +SR at mTESE, with an AUC of 70.3% (95% CI: 59.8–80.7). Decision curve analysis displayed the net clinical benefit of using an AMH <4 ng/ml threshold. LIMITATIONS, REASONS FOR CAUTION There is a need for external validation in even larger cohorts, across different centres and ethnicities. Systematic reviews and meta-analysis to provide high level of evidence are lacking in the context of AMH and SR rates in men with iNOA. WIDER IMPLICATIONS OF THE FINDINGS Current findings suggest that slightly more than one in two men with iNOA had −SR at mTESE. Overall, men with iNOA with lower levels of AMH had a significantly higher percentage of successful SR at surgery. A threshold of <4 ng/ml for circulating AMH ensured satisfactory sensitivity, specificity, and positive predictive values in the context of +SR at mTESE. STUDY FUNDING/COMPETING INTEREST(S) This work was supported by voluntary donations from the Urological Research Institute (URI). All authors declare no conflict of interest. TRIAL REGISTRATION NUMBER N/A