To assess the incidence of second brain tumours in patients operated and irradiated for pituitary tumours. The study base consisted of a consecutive series of 325 patients operated and irradiated for pituitary tumours, excluding patients with acromegaly and Cushing's disease. Comparison was made with the general population from the same catchment area as the patients. The follow-up period started in 1958 and on an individual basis patients were followed from the onset of postoperative irradiation until December 1995, or until date of death, emigration or a second brain tumour diagnosis, whichever occurred first. Three brain tumours (two astrocytomas and one meningioma) were observed, compared with 1-13 expected (standardized incidence ratios (SIR) 2.7; 95% confidence interval (CI) 0.6-7.8). The present study gives no firm support for an increased incidence of a second brain tumour in patients operated and irradiated for pituitary tumours. A crude meta-analysis of the present and previously published cohort studies of patients with irradiated pituitary tumours gives an SIR of 6.1 (95% CI 3.16-10.69). Thus, the results of the meta-analysis are in favour of an increased risk for second brain tumours. A genetic trait that predisposes to both pituitary tumours and brain tumours is an alternative causal factor. There is no definite proof that cranial irradiation per se is the causal factor. This question cannot be fully answered until sufficient cohort studies of nonirradiated pituitary tumour patients have been carried out.