The role of ultrasound as a screening test for choledocholithiasis was prospectively assessed by comparing the results of upper abdominal ultrasound with direct cholangiography in 59 unselected symptomatic postcholecystectomy patients. Ultrasound detected duct stones in 13 of 29 patients (sensitivity, 45%) and their absence in 29 of 30 (specificity, 97%). A positive ultrasound diagnosis of choledocholithiasis was correct 13 times out of 14 (predictive value, 93%) whereas a negative diagnosis was correct on only 29 of 45 occasions (predictive value, 64%). No significant learning effect was seen. Intestinal gas obscuring the distal common duct was the most important factor limiting the ability of ultrasound to detect duct stones. Duct stones were present in 25 of 35 patients shown to have a dilated common duct on ultrasound, and in 4 of 24 with nondilated ducts; the predictive value of duct dilatation at ultrasound for duct stones was therefore 71% and that of nondilatation in excluding stones was 83%. Dilated intrahepatic bile ducts were demonstrated at ultrasound in only 17% of patients with choledocholithiasis. We conclude that ultrasound cannot reliably diagnose or exclude bile duct stones and is an inadequate screening test for the symptomatic postcholecystectomy patient.