Orientia tsutsugamushi is the causative agent of scrub typhus. For the diagnosis of scrub typhus, we investigated the performances of conventional PCR (C-PCR), nested PCR (N-PCR), and real-time quantitative PCR (Q-PCR) targeting the O. tsutsugamushi-specific 47-kDa gene. To compare the detection sensitivities of the three techniques, we used two template systems that used plasmid DNA (plasmid detection sensitivity), including a partial region of the 47-kDa gene, and genomic DNA (genomic detection sensitivity) from a buffy coat sample of a single patient. The plasmid detection sensitivities of C-PCR, N-PCR, and Q-PCR were 5 × 10(4) copies/μl, 5 copies/μl, and 50 copies/μl, respectively. The results of C-PCR, N-PCR, and Q-PCR performed with undiluted genomic DNA were negative, positive, and positive, respectively. The genomic detection sensitivities of N-PCR and Q-PCR were 64-fold and 16-fold (crossing point [Cp], 37.7; 426 copies/μl), respectively. For relative quantification of O. tsutsugamushi bacteria per volume of whole blood, we performed real-time DNA PCR analysis of the human GAPDH gene, along with the O. tsutsugamushi 47-kDa gene. At a 16-fold dilution, the copy number and genomic equivalent (GE) of GAPDH were 1.1 × 10(5) copies/μl (Cp, 22.64) and 5.5 × 10(4) GEs/μl, respectively. Therefore, the relative concentration of O. tsutsugamushi at a 16-fold dilution was 0.0078 organism/one white blood cell (WBC) and 117 organisms/μl of whole blood, because the WBC count of the patient was 1.5 × 10(4) cells/μl of whole blood. The sensitivities of C-PCR, N-PCR, and Q-PCR performed with blood samples taken from patients within 4 weeks of onset of fever were 7.3% (95% confidence interval [CI], 1.6 to 19.9), 85.4% (95% CI, 70.8 to 94.4), and 82.9% (95% CI, 67.9 to 92.8), respectively. All evaluated assays were 100% specific for O. tsutsugamushi. In conclusion, given its combined sensitivity, specificity, and speed, Q-PCR is the preferred assay for the diagnosis of scrub typhus.