Benzodiazepines are often used in the treatment of cancer related breathlessness but there are no published trials of their use in these patients. Results from the few small trials conducted in patients with COPD are contradictory. A double-blind, randomised crossover trial comparing 5 days of lorazepam (0.5 mg 12 hourly) and placebo, with a 2 day washout period between, was conducted in breathless in- and out-patients with cancer who were not taking regular benzodiazepines. The primary outcome measure was change in VAS score for "trouble caused by breathlessness". Secondary outcome measures were change in VAS scores for "breathing at best" and "breathing at worst" and changes in HADS anxiety score. Patients scored the effectiveness of each intervention and, at the end of the study, indicated any preference for continuing either treatment. 26 patients were entered, of whom 17 completed the trial. Neither intervention resulted in a clinically significant improvement in symptoms. There was no statistical difference between interventions for any of the outcome measures (p-values 0.974, 0.159, 0.426 and 0.875 respectively). No correlation was found between any effect on breathlessness and either anxiety scores or pre-existing cardiorespiratory disease. Attributed side effects were infrequent; drowsiness was the most common, occurring in 4 patients on lorazepam and 2 on placebo. There was no statistically significant difference in effectiveness scores. 6 patients wished to continue lorazepam, 3 placebo and 3 either intervention. Despite the limitations of this small trial, the results do not support the use of regular benzodiazepines for cancer-related breathlessness.