We conducted a seroepidemiological survey of human plague in the general population using random sampling in the area of Ambositra, the main focus of plague in the central highlands of Madagascar (520 confirmed and presumptive cases notified during the past 10 years). Sera were tested using an ELISA IgG F1 assay. Considering the internal validity of the assay and the sampling method, the overall corrected prevalence of F1 antibodies was 0.6% (95% CI: 0.2%-1.8%). Being nearly 0 up to the age of 40, the corrected prevalence increased markedly after 45 years to 6.2%. Six of 20 individuals who declared to have been treated for clinical suspicion of bubonic plague in the past had F1 antibodies. The seroprevalence did not differ according to gender except in individuals > 60, where antibodies were significantly more frequent in males. This study suggests that the number of clinically suspected cases of plague provided by the surveillance network was plausible, despite some true cases being missed and a significant number of false positives. We also confirm that Yersinia pestis infections may occur without marked clinical manifestations and patients may recover without treatment, in accordance with old observations of pestis minor.