We report an open, uncontrolled study to evaluate the effectiveness of regular oral morphine as symptomatic treatment of dyspnoea in patients with advanced cancer receiving standard clinical care. Fifteen patients were assessed initially, and then 48 h and 7-10 days after starting treatment with oral morphine or having their dose increased. Dyspnoea, measured on a visual analogue scale (0-100), fell by a median of 14 (95% confidence interval -1.5, 25.5; Wilcoxon statistic 32.0; P = 0.06) in the nine who completed all three assessments. The three patients who died during the study did not show symptomatic benefit and, like the three who withdrew, experienced increased sedation and/or dizziness. Sedation was significantly increased at 48 h; median rise 10.5 (95% confidence interval 7, 25; Wilcoxon statistic 74; P = 0.007). Baseline respiratory function (FEV1, FVC, peak flow) was poor and the patients' respiratory rate was unaffected. Regular, titrated oral morphine may improve dyspnoea in some patients with advanced cancer but can cause significant short-term adverse effects. Oral morphine should be given to these patients as a therapeutic trial. Patients should be advised about side-effects and carefully monitored. Larger studies are needed to establish which patients are most likely to benefit and optimal dosage regimens.