BACKGROUND We aimed to determine pregnancy and miscarriage rates following intracytoplasmic sperm injection (ICSI) cycles using retrieved epididymal and testicular sperm in azoospermic men and ejaculated sperm in oligospermic and normospermic men. METHODS This retrospective study was carried out on 517 couples who underwent ICSI. They included 96 couples with azoospermia and 421 with oligospermia or normal sperm count in the male partner. Of the men with azoospermia, 69 underwent percutaneous epididymal aspiration (PESA) and 47 underwent testicular sperm extraction (TESE). In the 421 men with oligospermia or normal sperm count, ejaculated sperm was used for ICSI. The differences in the outcomes of ICSI using PESA or TESE and ejaculated sperm were evaluated. The main outcome measures were pregnancy and miscarriage rates. RESULTS No significant differences were seen in pregnancy and miscarriage rates with surgically retrieved and ejaculated sperm. The pregnancy rates (including frozen embryo transfer) were 43.5%, 36.2%, and 41.4% in couples with PESA, TESE, and ejaculated sperm, respectively (P = .93). The miscarriage rates were 16.7%, 23.5%, and 12.1%, respectively (P = .37). CONCLUSIONS Intracytoplasmic sperm injection in combination with PESA and TESE is an effective method and can successfully be performed to treat men with azoospermia. The outcomes with these procedures are comparable to ICSI using ejaculated sperm.