Abstract Study question To investigate the value of anti-Müllerian hormone(AMH) determination for estimating the sperm retrieval rate(SRR) of microdissection testicular sperm extraction(micro-TESE) in non-obstructive azoospermia(NOA) patients. Summary answer NOA with low AMH would have more opportunity to present heterogeneous seminiferous tubules when micro-TESE was performed and had higher SRR, especially in idiopathic cases. What is known already For infertile patients with NOA, micro-TESE is considered to have higher SRR than traditional surgery methods. However, serum inhibin B, follicle-stimulating hormone (FSH) and various clinical parameters are not reliable predictors for the presence of focal spermatogenesis and SRR. In male, AMH is a glycoprotein secreted by Sertoli cells and facilitate the regression of Müllerian structures in the developing foetus. It is still controversial that whether AMH level has value to predict the SRR of micro-TESE. Study design, size, duration This was a retrospective case-control study. From September 2014 to May 2022, 502 NOA patients treated with micro-TESE were divided into different groups according to their surgery outcome and seminiferous tubules appearance. Age, testis volume, serum AMH, FSH and testosterone level were compared between the different groups. The differences of SRR and AMH level in NOA patients with different etiologies were also compared. Participants/materials, setting, methods Micro-TESE was performed at x10 to x20 magnification. An attempt was made to identify seminiferous tubules that were larger and more opaque than others. The procedure was terminated when sperm were retrieved. If all tubules were seen to have an identical appearance, at least three samples (upper, middle, and lower) were extracted. Venous blood samples were drawn from each patient (7–10 AM) after an overnight fast. FSH and AMH were measured by electrochemiluminescence immunoassay. Main results and the role of chance Testicular sperms were successfully retrieved in 270 cases (SRR=53.8%). There were no statistical differences in age, testicular volume, FSH and testosterone levels between the patients who succeeded and failed to obtain sperm (all P>0. 05). The patients who obtained sperms had lower serum AMH level than those without sperm [0.81(0.16, 3.26)μg/L vs.1.37(0.21, 4.84)μg/L, P<0.05]. Patients with orchitis or AZFc deletion, cryptorchidism, KS, idiopathic azoospermia would have different AMH level [0.15(0.01, 0.41)μg/L, 5.71(3.57, 8.26)μg/L, 2.29(1.36, 3.81)μg/L, 0.15(0.05, 0.39)μg/L, 2.46(0.75, 5.49)μg/L, P<0.05]. Idiopathic azoospermia patients who obtained sperms had higher age but lower testosterone and AMH level than those without sperm [(35.2±8.9) years vs. (32.5±5.5) years, P<0.05, (3.1±1.4)μg/L vs. (3.7±2.1)μg/L, P<0.05; 1.63(0.35, 3.84)μg/L vs.3.00(1.20, 6.68)μg/L, P<0.05], there were no statistical differences in testicular volume and FSH level between the two groups (P>0.05). Receiver operating characteristic (ROC) curve showed that cut-off of serum AMH for successful sperm retrieval of idiopathic azoospermia patients was determined to be 2.96, with a sensitivity of 0.710 and specificity of 0.523, area under the curve (AUC) was 0.649. In the cases presenting heterogeneous seminiferous tubules during micro-TESE had lower AMH level and higher SRR than those presenting homogeneous seminiferous tubules [0.55(0.12, 2.05)μg/L vs.2.99(0.76, 6.11)μg/L, 75.9%(236/311)vs. 17.8%(34/191), all P<0.05]. Limitations, reasons for caution Pathology analysis should be involved in the following study. Randomized controlled trial comparing micro-TESE and traditional TESE would demonstrate that whether idiopathic azoospermia patients with higher serum AMH level would have less benefit by microsurgery than patients with lower AMH level. Wider implications of the findings Recently, in our pathologic research, NOA patients with extremely lower serum AMH level were observed to have more opportunity to present severe hyalinization in their seminiferous tubules. Tubules with severe hyalinization have less Sertoli cells and seem very thin. Therefore, tubules with spermatogenesis would be easy to identify during micro-TESE. Trial registration number not applicable