Background: Mansonella ozzardi is a poorly understood human filarial parasite with a broad distribution throughout Latin America. Most of what is known about its parasitism has come from epidemiological studies that have estimated parasite incidence using light microscopy. Light microscopy can, however, miss lighter, submicroscopic, infections. In this study we have compared M. ozzardi incidence estimates made using light microscopy, with estimates made using PCR. Methods: 214 DNA extracts made from Large Volume Venous Blood Samples (LVVBS) were taken from volunteers from two study sites in the Rio Solimões region: Codajás [n = 109] and Tefé [n = 105] and were subsequently assayed for M. ozzardi parasitism using a diagnostic PCR (Mo-dPCR). Peripheral finger-prick blood samples were taken from the same individuals and used for microscopic examination. Finger-prick blood, taken from individuals from Tefé, was also used for the creation of FTA®card dried blood spots (DBS) that were subsequently subjected to Mo-dPCR. Results: Overall M. ozzardi incidence estimates made with LVVBS PCRs were 1.8 times higher than those made using microscopy (44.9 % [96/214] compared with 24.3 % [52/214]) and 1.5 times higher than the PCR estimates made from FTA®card DBS (48/105 versus 31/105). PCR-based detection of FTA®card DBS proved 1.3 times more sensitive at diagnosing infections from peripheral blood samples than light microscopy did: detecting 24/105 compared with 31/105. PCR of LVVBS reported the fewest number of false negatives, detecting: 44 of 52 (84.6 %) individuals diagnosed by microscopy; 27 of 31 (87.1 %) of those diagnosed positive from DBSs and 17 out of 18 (94.4 %) of those diagnosed as positive by both alternative methodologies. Conclusions: In this study, Mo-dPCR of LVVBS was by far the most sensitive method of detecting M. ozzardi infections and detected submicroscopic infections. Mo-dPCR FTA®card DBS also provided a more sensitive test for M. ozzardi diagnosis than light microscopy based diagnosis did and thus in settings where only finger-prick assays can be carried-out, it may be a more reliable method of detection. Most existing M. ozzardi incidence estimates, which are often based on light microscope diagnosis, are likely to dramatically underestimate true M. ozzardi parasitism incidence levels.