Our aim was to describe patient flows in mandibular third molar surgery at oral and maxillofacial specialist units. Our hypothesis was that there are variations in how care is delivered and that the variations could be explained by inter-individual variations in surgeons' practice, the quality of the radiographs appended to the referral, and the staffing of the specialist units. A flow chart was constructed to simulate all possible patient flows in the care process. The chart begins with treatment planning, which was drawn up based on documents from the referring dentist or another caregiver; continues with the care process at the oral and maxillofacial surgery unit, including surgical consultations and radiological examinations; and ends with surgery. Surgeons at four oral and maxillofacial surgery units in the National Health Service in southern Sweden participated. The intention was to collect data on at least 100 patients who had undergone mandibular third molar surgery at each unit. Data on 361 patients were collected. The radiographs appended to the referral were judged to be inappropriate for the majority of the patients (61%). For 13% of these patients, supplementary radiographic examinations were made at the radiology clinic included in the unit, whilst 48% were examined at the oral and maxillofacial surgery clinic. There were eight different patient flow patterns. In one unit with three surgeons, eight different flow patterns were recorded, indicating an interindividual variation among the surgeons. In a second unit, six different flow patterns were recorded. In the last two units, the patient flows appeared to be the same at each unit, although the predominant patient flows in these two units differed. The number of patient visits to the specialist units ranged between one and three. In three specialist units, most patients were called twice whilst in one specialist unit most patients were called only once, to have the third molar removed. Differences existed in the care process. Overall, the number of patient visits seemed not to depend on whether the preoperative radiographic examination was judged to be appropriate or whether the additional radiographs were made at the radiology clinic.