Because of the difficulty in diagnosing acute cholecystitis in critically ill patients with severe intercurrent illness by clinical and imaging methods or percutaneous aspiration of the gallbladder, a trial of percutaneous cholecystostomy was performed in 24 patients in the intensive-care unit with persistent, unexplained sepsis after a complete clinical, laboratory, and radiologic search showed no alternative source of infection. Persistent high fevers, despite antibiotic therapy, were present in all patients, with elevated WBC count in 18 patients, vague abdominal tenderness in 11, and septic shock requiring vasopressors in 15. Sonographically, all patients had distended, spherical gallbladders, six had gallstones, eight had wall thickening, three had pericholecystic fluid, and four had Murphy's sign. All patients were seen by a senior abdominal surgeon, who agreed to a trial of percutaneous cholecystostomy. Fourteen patients (58%) responded to percutaneous cholecystostomy, as evidenced by a decrease in WBC count, defervescence, and the ability to be weaned off vasopressors. Bile cultures were positive in four patients. Ten patients (42%) did not respond to percutaneous cholecystostomy; five eventually died of unrelated causes. A respiratory source of infection was eventually found in three of these 10 patients, with no proved source of infection in the remainder. No complications related to catheter insertion occurred in this group of patients. Bile leaks occurred in two patients when the percutaneous cholecystostomy catheter was removed, but without serious consequence. Our experience suggests that a lower threshold for performing percutaneous cholecystostomy in this difficult clinical subset of patients is worthwhile.