OBJECTIVE To determine whether there is an association between parvovirus B19 infection and early spontaneous abortion at less than 20 weeks' gestation. METHODS Eighty samples of early spontaneous abortions were analyzed. Each sample was examined histologically for the presence of viral inclusions, and selected cases were analyzed for parvovirus using electron microscopy and in situ hybridization. Polymerase chain reaction DNA amplification for the virus was done in each case. Maternal sera were analyzed for immunoglobulin (Ig) M and IgG parvovirus antibodies and compared with temporally matched controls. RESULTS Five cases in the study group had evidence of seroconversion for parvovirus, compared with two controls. Products of conception from two of these five cases were positive for virus by polymerase chain reaction amplification, and only one of these two had a characteristic inclusion of parvovirus histologically. Conversely, five chorionic vesicles from mothers who had not seroconverted had histologic changes suggesting parvovirus infection, but all of these cases were negative for parvovirus using in situ hybridization, polymerase chain reaction, and electron microscopy. CONCLUSIONS Parvovirus B19 DNA was found in two of 80 early spontaneous abortuses. Although viral DNA was detected in two cases, there was no clear evidence that the infections caused fetal death. Neither case showed erythroblastosis with large numbers of inclusions, as is seen in hydropic fetuses with parvovirus infection. In addition, in five cases in which parvovirus infection was not documented serologically or by the polymerase chain reaction, there was erythroid nuclear clearing suggestive of parvovirus B19 inclusions. This indicates that histologic evaluation for parvoviral inclusions is not always reliable in early spontaneous abortuses.