Since 1992 two new nucleic acid-based tests (Gen-Probe Pace 2 and Abbott LCR) for the diagnosis of Neisseria gonorrhoeae infection have been approved by the U.S. Food and Drug Administration. We systematically assessed the quality of 21 studies that evaluated these tests' performance compared with that of culture for diagnosis of gonorrhea, on the basis of established criteria. We estimated overall test sensitivity and specificity by the testing method, sex, and anatomic site. None of the studies optimally fulfilled all quality criteria; few studies adequately used reference tests or described blinding. The sensitivity and specificity of nucleic acid hybridization (approximately 85%, approximately 98%) and amplification tests (approximately 95%, approximately 99%) were high and did not appear to differ substantially by sex or anatomic site. When proficiency in the performance of culture is high, the new tests are comparable to culture and may not offer a substantial advantage; in settings where optimization of culture is difficult, nucleic acid amplification may detect more infections than nucleic acid probe or culture.