Our objective was to establish which is the best sperm retrieval technique in non-obstructive azoospermia based on the available evidence. To date, no randomized controlled trial has compared the efficiency of these strategies and thus current recommendations are based on cumulative evidence provided by descriptive, observational and controlled studies. Three outcome measures were assessed for the sperm retrieval techniques: sperm retrieval rate (SRR), complications and live birth rate. Twenty-four descriptive studies reporting on the results of testicular sperm extraction (TESE) were encountered. Seven controlled studies that compared microdissection (MD) TESE with conventional TESE and seven controlled studies comparing fine needle testicular aspiration (FNA) with TESE were identified. The mean SRR for TESE was 49.5% (95% CI 49.0-49.9). TESE with multiple biopsies results in a higher SRR than FNA especially in cases of Sertoli-cell-only (SCO) syndrome and maturation arrest. Current evidence suggests that MD performs better than conventional TESE only in cases of SCO where tubules containing active focus of spermatogenesis can be identified. MD appears to be the safest technique regarding post-operative complications followed by FNA. Only three studies could be identified concerning the influence of the sperm retrieval technique on clinical pregnancy and live birth rate, hence no definitive conclusions can be made. However, so far there appears to be no impact of the technique itself on success rates.