The objective of the current study was to determine the sensitivity and specificity of real-time polymerase chain reaction (real-time PCR) for feline hemoplasmas when applied to DNA extracted from dried whole-blood smears in comparison to that for DNA extracted from liquid whole blood. Blood samples were collected into ethylenediamine tetra-acetic acid tubes from 305 cats with possible or suspected hemoplasmosis, and dried blood smears from each sample were prepared. DNA was extracted from blood smears and a 160-microl aliquot of each liquid blood sample by using a robotic extractor and was subjected to real-time PCR for feline glyceraldehyde-3-phosphate dehydrogenase (liquid blood), 18S ribosomal RNA (dried blood), and "Candidatus Mycoplasma haemominutum", Mycoplasma haemofelis, and "Candidatus Mycoplasma turicensis" DNA. When using the results for liquid whole blood as the gold standard, the sensitivity of each assay for "Ca. M. haemominutum", M. haemofelis, and "Ca. M. turicensis" was 49 of 66 (74%), 11 of 13 (85%), and 11 of 20 (55%), respectively. The specificity of each assay was 224 of 234 (96%), 287 of 287 (100%), and 280 of 280 (100%), respectively. When possible, liquid blood samples should be submitted for detection of feline hemoplasmas by using real-time PCR. The improved sensitivity of real-time PCR on blood smears for M. haemofelis compared with that of the other hemoplasma species may reflect the higher organism burdens associated with infection with this species.