Testicular sperm extraction (TESE) is often an effective method for sperm retrieval from men with non-obstructive azoospermia. However, TESE has been a blind procedure that does not identify the focal sperm-producing areas of the testicle until after tissue has been excised from the patient. Experience with a new technique of microdissection of testicular tubules is presented here that identifies sperm-containing regions before their removal. Identification of spermatogenically active regions of the testicle is possible by direct examination of the individual seminiferous tubules. The underlying concept for this technique is simple: seminiferous tubules containing many developing germ cells, rather than Sertoli cells alone, are likely to be larger and more opaque than tubules without sperm production. In a sequential series of TESE cases for men with non-obstructive azoospermia, the ability to find spermatozoa increased from 45% (10/22) to 63% (17/27) after introduction of the microdissection technique. Microdissected samples yielded an average of 160,000 spermatozoa per sample in only 9.4 mg of tissue, whereas only 64,000 spermatozoa were found in standard biopsy samples that averaged 720 mg in weight (P < 0.05 for all comparisons). For men where microdissection was attempted, successful identification of enlarged tubules was possible in 56% (15/27) of cases. However, spermatozoa were retrieved with microdissection TESE for six men in whom sperm retrieval was unsuccessful with standard TESE approaches (35% of all men with spermatozoa retrieved). These findings suggest that microdissection TESE can improve sperm retrieval for men with non-obstructive azoospermia over that achieved with previously described biopsy techniques.