Objective: The purpose of this study was to evaluate outcome of intracytoplasmic sperm injection (ICSI) using epididymal and testicular sperm in patients with azoospermia. Methods: From March, 1993 to May, 1999, a retrospective clinical analysis was done of a total of 140 cycles in 112 patients who underwent ICSI. Subjects were divided into three groups: ejaculated-ICSI group included 42 cycles in 34 patients with ejaculated sperm who underwent ICSI due to severe oligospermia and past history of failed or poor fertilization in the previous in vitro fertilization and embryo tranfer (IVF-ET) cycles, microsurgical epididymal sperm aspiration and intracytoplasmic sperm injection (MESA-ICSI) group included 50 cycles in 42 patients with congenital absence of the vas deferens (CAVD) or unreconstructable obstructive azoospermia and testicular sperm extraction and intracytoplasmic sperm injection (TESE-ICSI) group included 48 cycles in 36 patients with no spermatozoa which can be retrieved from epididymis or non-obstructive azoospermia. Results: Normal two-pronuclear fertilization rates were similar in three groups: 64.4% for ejaculated-ICSI group, 59.4% for MESA-ICSI group and 60.4% for TESE-ICSI group. The pregnancy rates were 26.2%, 26.0% and 25.0% respectively. There were no significant differences in the fertilization, cleavage, and clinical pregnancy rates among ICSI cycles using ejaculated, epididymal and testicular sperm. Conclusion: Epididymal and testicular sperm obtained in azoospermic patients can fertilize oocyte successfully and may lead to be similar fertilization rates and clinical pregnancy rates to ejaculated sperm.