The effect on transcutaneous SaO2 and transcutaneous carbon dioxide tension (PtCO2) of eating was assessed in 44 patients with severe COPD (FEV1 less than 50 percent). The SaO2, PtCO2, and heart rate (HR) were measured every minute before, during, and until 30 minutes after a standardized meal (445 kcal) was consumed. All patients were measured twice on the same day, while eating a meal with high (80 percent) and low (28 percent) carbohydrate content, respectively. The mean meal desaturation (delta SaO2) was less than 1 percent in normoxemic patients but was -3.2 +/- 0.7 percent (p less than 0.001) in hypoxemic (PaO2 less than 7.3 kPa) patients. Significant differences between hypoxemic patients with a delta SaO2 greater than 4 percent and less than or equal to 4 percent, respectively, were found in FEV1 (16 +/- 3 percent and 29 +/- 8 percent; p less than 0.001), respiratory muscle strength (3.9 +/- 1.2 kPa and 5.9 +/- 1.2 kPa; p less than 0.01), HR (112 +/- 12 beats per minute and 90 +/- 18 beats per minute; p less than 0.001), body weight (54.9 +/- 7.5 kg and 74.7 +/- 10.4 kg; p less than 0.001), and fat-free mass (42.0 +/- 6.6 kg and 52.6 +/- 5.8 kg; p less than 0.005) but not in baseline SaO2 and PtCO2. The decrease in SaO2 and the increase in HR were less during the carbohydrate-rich meal. No significant fluctuations in PtCO2 were found after either meal. Meal-related oxygen desaturation cannot explain weight loss in normoxemic patients with COPD but may contribute to a limited dietary intake in a subgroup of hypoxemic patients exhibiting marked oxygen desaturation during meals. A single carbohydrate-rich meal does not have an immediate impact on PtCO2 in stable COPD.