ABSTRACT Objective To determine the optimal cut-off value for follicle stimulating hormone (FSH) to predict the outcome of microsurgical testicular sperm extraction (micro-TESE) in patients with nonobstructive azoospermia (NOA). Subjects and methods We included a total number of 180 patients with NOA. The serum level of FSH was determined and all the subjects underwent micro-TESE. We determined the optimal cut-off value for FSH and assessed whether the test could be effectively used as a successful predictor of sperm retrieval by calculating the Receiver Operating Characteristic (ROC) area under the curve. Results Overall we included a total number of 171 patients with mean age of 34.3 ± 8.6 years. The micro-TESE was considered to be successful in 79 (43.8%) while it failed in 92 (56.2%) patients. We found that the mean level of serum FSH was significantly higher in group those with failed micro-TEST compared to successful group (p < 0.001). The cut-off value for FSH was calculated to be 14.6 mIU/mL to predictive the outcome of micro-TESE with a sensitivity of 83.5% [73.5%-90.9%] and a specificity of 80.3% [69.5%-88.5%]. At this value, the other parameters were calculated to be PPV, 81.5%; NPV, 82.4; LR+, 4.23; and LR-, 0.21. Conclusions The results of the current study indicate that FSH plasma levels above 14.6 mIU/mL can be considered to be the failure predictor of the micro-TESE in NOA patients.