The diagnostic value of a PCR assay that amplifies a 501-bp fragment of the Yersinia pestis caf1 gene has been determined in a reference laboratory with 218 bubo aspirates collected from patients with clinically suspected plague managed in a regional hospital in Madagascar. The culture of Y. pestis and the detection of the F1 antigen (Ag) by enzyme-linked immunosorbent assay (ELISA) were used as reference diagnostic methods. The sensitivity of PCR was 89% (57 of 64) for the Y. pestis-positive patients, and 80.7% (63 of 78) for the F1 Ag-positive patients. The specificity of PCR for the culture-, F1 Ag-, and antibody-negative patients (n = 105) was 100%. Because in Madagascar most patients with plague are managed and their clinical samples are collected in remote villages, the usefulness of PCR was evaluated for routine diagnostic use in the operational conditions of the control program. The sensitivity of PCR was 50% (25 of 50) relative to the results of culture and 35.2% (19 of 54) relative to the results of the F1 Ag immunocapture ELISA. The specificity of PCR under these conditions was 96%. In conclusion, the PCR method was found to be very specific but not as sensitive as culture or the F1 Ag detection method. The limitation in sensitivity may have been due to suboptimal field conditions and the small volumes of samples used for DNA extraction. This technique is not recommended as a routine diagnostic test for plague in Madagascar.