The effect of several pre- and per-operative variables on indicators of postoperative morbidity was assessed in 204 patients after unilateral mandibular third-molar surgery. The variables included gender, age, use of tobacco and alcohol, state of eruption, depth and angulation of the tooth, duration of the operation, pericoronitis, and time of day of surgery. Visual analogue scales were used for patient assessment of pain and swelling and for clinical assessment of swelling. Maximum pain was indicated 6 h postoperatively and maximum swelling the first postoperative evening. The results showed a mean reduction of mouth opening capacity (trismus) of 31% the 1st postoperative day. Mean analgesic consumption was 3.7 tablets, mean number of days of inability to work 1.1, and the rate of postoperative alveolitis 1.9%. The variation of the morbidity indicators was considerable. Multiple classification analyses showed that the predictors explained from 17% (clinically assessed swelling) to 8% (pain 6 h postoperatively) of the variance of these indicators. It is concluded that these commonly used predictors only to a minor extent can explain the wide variation in postoperative morbidity after mandibular third-molar surgery.