Treponema pallidum PCR (Tp-PCR) testing now is recommended as a valid tool for the diagnosis of primary or secondary syphilis. The objectives were to systematically review and determine the optimal specific target gene to be used for Tp-PCR. Comparisons of the performance of the two main targets are tpp47 and polA genes were done using meta-analysis. Three electronic bibliographic databases, representing abstract books from five conferences specialized in infectious diseases from January 1990 to March 2015, were searched. Search keywords included ("syphilis" OR "Treponema pallidum" OR "neurosyphilis") AND ("PCR" OR "PCR" OR "molecular amplification"). We included diagnostic studies assessing the performance of Tp-PCR targeting tpp47 (tpp47-Tp-PCR) or the polA gene (polA-Tp-PCR) in ulcers from early syphilis. All studies were assessed against quality criteria using the QUADAS-2 tool. Of 37 studies identified, 62.2% were judged at low risk of bias or applicability. Most used the U.S. Centers for Disease Control and Prevention (CDC) case definitions for primary or secondary (early) syphilis (89.2%; n = 33); 15 (40.5%) used darkfield microscopy (DFM). We did not find differences in sensitivity and specificity between the two Tp-PCR methods in the subgroup of studies using adequate reference tests. Among studies using DFM as the reference test, sensitivities were 79.8% (95% confidence intervals [CI], 72.7 to 85.4%) and 71.4% (46.0 to 88.0%) for tpp47-Tp-PCR and polA-Tp-PCR (P = 0.217), respectively; respective specificities were 95.3% (93.5 to 96.6%) and 93.7% (91.8 to 95.2%) (P = 0.304). Our findings suggest that the two Tp-PCR methods have similar accuracy and could be used interchangeably.