Human granulocytic ehrlichiosis (HGE) is usually diagnosed by immunofluorescent antibody (IFA) serology with Ehrlichia equi-infected neutrophils or HGE agent-infected cultured HL60 cells. The HGE agent and E. equi are antigenically diverse, and interpretation of serologic results is also often variable. Thus, we investigated the sensitivity and specificity of various HGE agent and E. equi antigens used for IFA diagnosis by three different laboratories. Serum samples from 28 patients with well- characterized HGE and 9 patients with suspected HGE who were investigated by PCR, blood smear examinations, and serology were used, along with 9 serum samples from patients with other rickettsial and ehrlichial infections. Each serum sample was tested with up to 10 different antigen preparations. Overall, qualitative IFA results agreed in 70% of the samples. Titers among antigens were similar (r = 0.89 to 0.96), but titers of individual samples varied by fourfold or more in 5 of 81 (6%) of the serum samples. Sensitivity ranged from 100% to 82%, and specificity varied from 100% to 67%, but these differences were not significant, even among those tested in the same laboratory or between two different laboratories. Antibodies were detected in 14 to 44% of acute-phase sera from confirmed HGE patients. Most false- positive reactions resulted with Ehrlichia chaffeensis; when these sera were excluded, the specificity of most antigens was 91 to 100%. These data indicate that IFA results often agree and that IFA is useful for diagnosis of HGE in convalescence. However, without further standardization, variability among serologic tests using E. equi and HGE agent isolates for diagnosis of HGE will occasionally provide discrepant results and confound diagnosis.