Modified logistic regression analysis of 24 variables in 300 patients undergoing laparoscopic cholecystectomy found the following parameters independently predictive for a difficult operation: right upper quadrant pain (p < 0.01), rigidity in right upper abdomen (p < 0.01), previous upper abdominal surgery (p < 0.01), biliary colic within the last 3 weeks (p < 0.05), white blood cell count > 10 x 10(9)/l (p < 0.05), thickening of the gallbladder wall (p < 0.05), hydroptic gallbladder (p < 0.05), pericholecystic fluid (p < 0.01), shrunken gallbladder (p < 0.01), and no filling of the gallbladder in preoperative intravenous cholangiography (p < 0.05). Based on these variables a diagnostic model was developed to predict the difficulty of a laparoscopic cholecystectomy, with scores ranging from 0 (ideal case) to IV (conversion to open cholecystectomy expected) prior to surgery. When the reliability of our model was examined in a second study in 340 consecutive patients undergoing laparoscopic cholecystectomy 80% of the patients were predicted correctly. Our model should help to select patients for either laparoscopic or open cholecystectomy based on the expected difficulties and the experience of the surgeon.