Molecular pathology may demonstrate tumour cells not detected by histology. The idea has emerged that these cells influence the prognosis negatively and that their detection will lead to more appropriate treatment and improved patient survival. We theorized that tumour cells at surgical margins overlooked by the pathologist should demonstrate their clinical significance by causing recurrences at the primary site in the patients reported to have tumour-free margins by histology. To assess this assumption, we investigated the prognostic influence of the histologically determined status of the surgical margins. The material that formed the basis of this study consisted of 394 patients that underwent resection for their primary tumour during the years 1990-1995. In 207 patients, initial treatment was complete as assessed by conventional histopathological examination of the surgical specimen. In 187 patients, initial treatment was incomplete, defined as tumour in or close to the margin, or mild, moderate or severe dysplasia or in situ cancer at the margin. Causes for treatment failure were recorded for both groups separately. In the group with tumour-free margins, 16.9% had a second primary head and neck cancer, 8.2% had a second tumour in the lung, 10.6% had recurrent disease in the neck, 2.9% had distant metastasis, and 3.9% had local recurrence at the same site as the primary cancer. For the group without tumour-free margins, these figures were the following: second primary in the head and neck area: 17.1%, second primary in the lung: 7.0%, recurrent disease in the neck: 11.8%, distant metastasis: 8.0% and local recurrence at the primary site: 21.9%. Local recurrences were rare in patients in which the pathologist reported the resection to be complete. Although there may be tumour cells in surgical margins that evade histological detection, their clinical impact appears to be almost negligible.