Based on the neurophysiology of dyspnoea and the distribution of cannabinoid receptors within the central nervous system, we hypothesize that the unpleasantness of breathlessness will be ameliorated in humans by cannabinoids, without respiratory depression. Five normal and four chronic obstructive pulmonary disease (COPD) subjects entered a double blind, randomized, placebo-controlled crossover study with two test days. Subjects received sublingual cannabis extract or placebo. A maximum of 10.8 mg tetrahydrocannabinol and 10 mg cannabidiol were given. Breathlessness was simulated using fixed carbon dioxide loads. Measurements taken were of breathlessness (visual analogue scale [VAS] and breathlessness descriptors), mood and activation, end-tidal carbon dioxide tension and ventilatory parameters. These were measured at baseline and 2 hours post placebo and drug administration. Normal and COPD subjects showed no differences in breathlessness VAS scores and respiratory measurements before and after placebo or drug. After drug administration, COPD subjects picked 'air hunger' breathlessness descriptors less frequently compared to placebo. We have shown that breathlessness descriptors may detect an amelioration of the unpleasantness of breathlessness by cannabinoids without a change in conventional breathlessness ratings (VAS). A stimulus more specific for air hunger may be needed to demonstrate directly a drug effect on breathlessness. However, this study shows that the inclusion of respiratory descriptors may contribute to the assessment of drug effects on breathlessness.