The effects of sample storage, target preparation and annealing temperature on a nested polymerase chain reaction (PCR) test for Toxoplasma gondii DNA were investigated with experimentally seeded amniotic fluids to simulate congenital infection. Aliquots of 17 amniotic fluid samples remaining after investigation for conditions other than toxoplasmosis and seeded to contain small numbers of T. gondii tachyzoites, were tested after storage for up to 2 weeks. There was no deterioration in test sensitivity on samples stored for 1-2 weeks at room temperature; thus, samples sent by post to reference laboratories are acceptable for examination. There was no significant difference in the results from two target preparation methods or two different annealing temperatures. One-to-ten parasites were detectable in 13 of 17 test samples. The significant feature in the remaining four samples was the use of washed stored parasites to seed the amniotic fluids rather than any feature of the amniotic fluids used or the test. False positive results were found in up to 15% of unseeded amniotic fluid samples, which contrasts with only 1.9% for the same PCR test in other routine or negative control samples tested as part of the laboratory's reference diagnostic work. The difference illustrates the increased risk of cross-contamination in PCR when the majority of specimens tested are positive. The results suggest that PCR techniques are likely to be sensitive and effective tools for the routine diagnosis of congenital toxoplasma infection.