Human granulocytic ehrlichiosis is an emerging infectious disease in the United States and Europe, and PCR methods have been shown to be effective for the diagnosis of acute infections. Numerous PCR assays and primer sets have been reported in the literature. The analytical sensitivities (limits of detection) of 13 published PCR primer sets were compared using DNA extracted from serial dilutions of Anaplasma phagocytophilum-infected HL-60 cells. The specificity of the assays that were able to detect <or=2.5 infected cells was tested by the use of template DNA extracted from Ehrlichia chaffeensis, Rickettsia rickettsii, and Bartonella henselae. The assays with the lowest limits of detection were shown to be a nested assay that amplifies the 16S rRNA gene (primer pairs ge3a-ge10 [primary] and ge9-ge3 [nested]; detects 0.25 infected cell), a direct assay that amplifies the major surface protein gene msp2 (primer pair msp2-3f-msp2-3r; detects 0.25 infected cell), and a direct assay that amplifies the 16S rRNA gene (primer pair ehr521-ehr790; detects 0.25 infected cell). The specificity and limit of detection of the MSP2 and 16S rRNA direct assays were further tested by use of A. phagocytophilum template DNA from both North America and Europe and from human, tick, white-footed mouse, equine, deer, bovine, and wood rat samples and of template DNA from closely related species (Anaplasma marginale, the white-tailed deer agent, and additional E. chaffeensis-positive samples). Three manufacturers' PCR kits were tested and showed distinct variations in the limit of detection, specificity, and nonspecific background amplification. The importance of these results for the molecular diagnosis of human granulocytic ehrlichiosis is discussed.