Purpose The purpose of this study was to evaluate the best protocol to prepare endometrium for frozen embryo replacement (FER) cycles. Methods This study is a systematic review and meta-analysis. Following PubMed and OvidSP search, a total of 1166 studies published after 1990 were identified following removal of duplicates. Following exclusion of studies not matching our inclusion criteria, a total of 33 studies were analyzed. Primary outcome measure was live birth. The following protocols, including true natural cycle (tNC), modified natural cycle (mNC), artificial cycle (AC) with or without suppression, and mild ovarian stimulation (OS) with gonadotropin (Gn) or aromatase inhibitor (AI), were compared. Results No statistically significant difference for both clinical pregnancy and live birth was noted between tNC and mNC groups. When tNC and AC without suppression groups are compared, there was a statistically significant difference in clinical pregnancy rate in favor of tNC, whereas it failed to reach statistical significance for live birth. When tNC and AC with suppression groups are compared, there was a statistically significant difference in live birth rate favoring the latter. Similar pregnancy outcome was noted among mNC versus AC with or without suppression groups. Similarly, no difference in clinical pregnancy and live birth was noted when ACs with or without suppression groups are compared. Conclusions There is no consistent superiority of any endometrial preparation for FER. However, mNC has several advantages (being patient-friendly; yielding at least equivalent or better pregnancy rates when compared with tNC and AC with or without suppression; may not require LPS). Mild OS with Gn or AI may be promising.