We and others have noted that desaturation occurs relatively soon after the onset of eating . This mitigates against explaining the changes on the basis of the metabolic effects of food, changes that would require time for digestion and absorption. Several authors have postulated that ventilation perfusion relationships are altered if tidal volume or respiratory frequency is altered during eating, and this alteration may be important in COPD . However, at least in normal subjects eating and drinking do not appear to affect significantly tidal volume, inspiratory duration, expiratory duration or minute ventilation, although breathing becomes more irregular . Finally, it is likely that eating, although usually considered a sedentary activity, is a low intensity exercise in patients with severe COPD. The muscles involved are those of mastication, swallowing and, of course, arm elevation. Couser et al have shown that simple arm elevation in normal subjects is associated with a significant increase in metabolic rate, with increased oxygen uptake and carbon dioxide production. This may at least partially explain the finding of oxygen desaturation during eating and the greater effect of AE than PE.