OBJECTIVE To compare amniotic fluid index (AFI) with the single deepest pocket in the identification of actual abnormal amniotic fluid (AF) volumes. METHODS One hundred seventy-nine women with singleton pregnancies at the University of Mississippi between March 1994 and June 1999 had ultrasound estimations of AF volume sequentially using the AFI and single deepest pocket techniques. Each woman subsequently had ultrasound-directed amniocentesis with dye-dilution and spectrophotometric calculation of actual AF volume. RESULTS Actual AF volumes were low (under 5% by volume for gestational age) in 62 women, normal (5-95%) in 100 women, and high (more than 95%) in 17 women. An AFI up to 5 cm (sensitivity 10%, specificity 96%) and a single deepest pocket up to 2 cm (sensitivity 5%, specificity 98%) were similarly inadequate in identifying dye-determined low AF volumes. Likewise, AFI above 20 (sensitivity 29%, specificity 97%) and a single-deepest pocket above 8 cm (sensitivity 29%, specificity 94%) were poor in identifying dye-determined abnormally high volumes. CONCLUSIONS There was no difference between AFI and single deepest pocket techniques for identifying truly abnormal AF volumes. Both techniques were unreliable for identifying true AF volumes.