Exit-site infections (ESI) and/or tunnel infections (TI) are common causes of peritoneal catheter loss. Catheter-related infection can be characterized by ultrasonographic examination (UE) of the exit site and the tunnel. This study was done to determine if the UE is useful in predicting catheter loss. Records from 1990 to 1992 of all continuous ambulatory peritoneal dialysis (CAPD) patients with an ESI or TI who had an UE performed were reviewed for causative organism, loss of catheter, and infection resolution. ESI was defined as redness or induration of exit site with or without the presence of drainage. TI was defined as induration or redness along tunnel or incision site with or without the presence of drainage. The presence of a pericatheter sonolucent fluid collection was considered a positive (+) UE, and the absence of fluid a negative (-) UE. Eleven CAPD patients had 12 UE performed within a week of starting antibiotic therapy. Ten were (+) and 2 (-); 7 catheters were removed, all with (+) UE. Of those removed, 6 were due to S. aureus, and 1 due to mycobacterial infection. Of the 3 (+) catheters not removed, 2 infections resolved clinically: both reverted to (-) UE after treatment. We conclude that a (+) UE is a useful predictor of catheter loss.