Clinical data from 259 patients with suspected acute appendicitis were prospectively collected and used in construction of a Bayesian scoring system, comprising 19 attributes, for preoperative diagnosis. The scoring system was integrated into the routine clinical management of a prospective series of 830 patients. Laparotomy was performed in 310 cases, and of these 256 had acute appendicitis (perforation in 14%). Excluding 6 cases with normal appendix but laparotomy mandatory for other reasons, the negative laparotomy rate was 15.5%. This rate was significantly lower than in earlier series from the same hospital and in more than 8,000 appendectomies performed in Sweden since 1969. The scoring system had 90.2% sensitivity, 91.4% specificity, 82.5% positive predictive value and 95.4% negative predictive value. The system is regarded only as an aid in diagnosis, to be used in combination with clinicians' judgements. Construction of a local data base probably is essential for results equivalent to those here reported.