Extracts of amniocentesis samples from 144 women were tested for the presence of mutagenic substances using tester strain TA1538 in the Ames Salmonella/mammalian-microsome mutagenicity test. Because the volume of amniotic fluid in these samples was limited (generally less than 10 ml), we investigated modifications of this mutagenesis assay that could increase its ability to detect effects from small quantities of test material. Using mutagenicity in samples of urine from smokers as a model, it appeared that improved ability to detect small amounts of mutagen could be obtained by reducing volumes of media and reagents while keeping the amount of test sample constant. This modification resulted in a test procedure capable of readily detecting mutagenicity in volumes of urine from smokers that were smaller than the volumes of amniotic fluid available. Tests of amniotic fluid extracts by this modified procedure showed small increases in revertants, about 50% above dimethylsulfoxide solvent control values. Results of procedures to control for technical factors possibly contributing to these increases suggested that the increased values could not be readily explained by contamination of test samples with mutagens during the extraction procedure. They also were not explained by alterations in spontaneous numbers of revertants associated with changes in the density of bacterial lawn growth. The increases suggest the presence of small amounts of mutagenic material in many of the amniotic fluid samples. At the doses employed, mutagenic activity in these samples was not associated with maternal smoking.