The rate of parvovirus B19 (PV) infection in cases of "idiopathic" nonimmune hydrops fetalis (NIHF) is reported to be approximately 16% with polymerase chain reaction (PCR)-based methods. Antibodies for use in paraffin-embedded tissue have not been systematically compared with PCR or with the presence of inclusions at varying gestational ages. All autopsy cases of NIHF and those with effusions of multiple serous membranes examined between 1991 and 1996 (n = 29) were evaluated for the presence of PV DNA by PCR analysis of paraffin-embedded liver tissue. PCR-positive cases and "idiopathic" cases were examined for the presence of inclusions in routine histological sections and for PV protein using a monoclonal antibody (NovoCastra R92F6). Among the four clinically idiopathic cases, one (25%) was positive for PV using PCR. The three negative idiopathic cases had no inclusions and were negative for PV by PCR and immunohistochemistry (IHC); all were third-trimester gestations (28, 31, and 32 weeks). Identifiable risk factors for NIHF other than PV in the remaining 25 cases included cystic hygroma, seven (three 45,X; two 46,XX; two no growth); complex cardiac anomaly, six; infection, three (two CMV, one chlamydia); twin-twin transfusion, two; lymphangiectasia, two; diaphragmatic hernia, tracheal atresia, trisomy 21, congenital cystic adenomatoid malformation, one each. One of these nonidiopathic cases, a fetus with cystic hygroma and a 45,X karyotype, was positive for PV DNA only on the blot, consistent with a low titer; no inclusions were present, and IHC was negative in multiple organs in this instance. One of four (25%) cases of idiopathic NIHF cases contained PV DNA by PCR analysis; there were abundant inclusions in multiple organs, and IHC was strongly positive as well. Of 25 cases of nonidiopathic NIHF, one (4%) was also positive for PV DNA by PCR. PV protein was detected by IHC only in the presence of inclusions; IHC thus may be useful for highlighting sparse inclusions. No second-trimester case of NIHF was unexplained. Late (third-trimester) cases of "idiopathic" NIHF are likely to be negative by all methods, either because they are not attributable to PV infection or because PV protein and DNA are below detectable levels or are no longer present. Maternal serology for PV and TORCH agents may be the best method for investigating third-trimester losses to otherwise unexplained NIHF.