Four hundred and six patients with primary non-small cell carcinoma of the bronchus causing symptoms due to endobronchial disease, were treated with intraluminal radiotherapy (ILT) using the microSelectron-HDR machine at the Christie Hospital, Manchester, between April 1988 and the end of 1992. An assessment of morbidity for this treatment is presented, particularly with regard to the risk factors and causes of massive haemoptysis death. The most common early side-effect was a mild transient exacerbation of cough which usually resolved within 2-3 weeks. At various times following ILT treatment 83 bronchoscopies were carried out randomly in 55 patients. In bronchoscopies carried out within the first 3 months following ILT, no tumour was visible in 80% of cases. A mucosal radiation reaction score (RRS) was used to grade bronchoscopic appearance after ILT treatment. Overall, 55% of bronchoscopic examinations showed some degree of mucosal radiation reaction. The majority of radiation reactions from 6 months onwards after ILT demonstrated a degree of fibrosis. A radiation reaction was seen more frequently after treatment with 2000 cGy as opposed to 1500 cGy at 1 cm from the central axis of the radiation source. Thirty-two patients were identified who had died from massive haemoptysis (MH) as a terminal event. A Cox multivariate regression analysis showed that the treatment-related factors of increased dose at first ILT (P = 0.004), prior laser treatment at the site of ILT (P = 0.020) and second ILT treatment in the same location as the first ILT treatment (P = 0.047), all significantly increased the relative risk of MH death compared with their effect on the relative risk of death from other causes (OC). (In addition a fourth treatment-related factor, namely the concurrent use of ILT and external beam radiotherapy (EB) had a P value of 0.08). Twenty out of 25 assessable MH-death patients (80%) had evidence of recurrent or residual tumour before death but 5 patients (20%) did not. For surviving patients the instantancious risk of death at any one time (the cause-specific death rate expressed as deaths per 100 cases per month), showed a sharp peak for MH deaths between 9 and 12 months post ILT in contradistinction to OC death where the peak was between 3 and 6 months post ILT. These findings may imply a role for late radiation reaction in the treatment-related risk factors identified as increasing the relative risk of MH death and possible mechanisms are discussed. The results have implications for treatment regimes that use a dose of 2000 cGy at 1 cm in a single fraction technique, that have a high frequency of previous laser treatment, that use multiple, repeated ILT treatments in the same location and that use ILT concurrently with EB.