Background Reference diagnostic tests for leptospirosis include nucleic acid amplification tests, bacterial culture, and microscopic agglutination testing (MAT) of acute and convalescent serum. However, clinical laboratories often do not receive paired specimens. In the current study, we tested serum samples using a highly sensitive real-time nucleic acid amplification test for Leptospira and compared results to MAT performed on the same specimens. Methods/Principal Findings 478 serum samples from suspected leptospirosis cases in Rio de Janeiro were tested using a real-time RT-PCR for the diagnosis of leptospirosis, malaria and dengue (the Lepto-MD assay). The Lepto-MD assay detects all species of Leptospira (saprophytic, intermediate, and pathogenic), and in the current study, we demonstrate that this assay amplifies both Leptospira RNA and DNA. Dengue virus RNA was identified in 10 patients, and no cases of malaria were detected. A total of 65 samples (13.6%) were positive for Leptospira: 35 samples (7.3%) in the Lepto-MD assay, 33 samples (6.9%) by MAT, and 3 samples tested positive by both (kappa statistic 0.02). Poor agreement between methods was consistent regardless of the titer used to define positive MAT results or the day of disease at sample collection. Leptospira nucleic acids were detected in the Lepto-MD assay as late as day 22, and cycle threshold values did not differ based on the day of disease. When Lepto-MD assay results were added to the MAT results for all patients in 2008 (n=818), the number of detected leptospirosis cases increased by 30.4%, from 102 (12.5%) to 133 (16.3%). Conclusions/Significance This study demonstrates a lack of agreement between nucleic acid detection of Leptospira and single-specimen MAT, which may result from the clearance of bacteremia coinciding with the appearance of agglutinating antibodies. A combined testing strategy for acute leptospirosis, including molecular and serologic testing, appears necessary to maximize case detection.