In animals, glucose concentrations are 3-20 times lower in lung lining fluid than in plasma. In humans, glucose concentrations are normally low (<1 mmol/l) in nasal and bronchial fluid, but they are elevated by inflammation or hyperglycemia. Furthermore, elevated bronchial glucose is associated with increased respiratory infection in intensive care patients. Our aims were to estimate normal glucose concentrations in fluid from distal human lung sampled noninvasively and to determine effects of hyperglycemia and lung disease on lung glucose concentrations. Respiratory fluid was sampled as exhaled breath condensate, and glucose was measured by chromatography with pulsed amperometric detection. Dilution corrections, based on conductivity, were applied to estimate respiratory fluid glucose concentrations (breath glucose). We found that breath glucose in healthy volunteers was 0.40 mmol/l (SD 0.24), reproducible, and unaffected by changes in salivary glucose. Breath-to-blood glucose ratio (BBGR) was 0.08 (SD 0.05). Breath glucose increased during experimental hyperglycemia (P < 0.05) and was elevated in diabetic patients without lung disease [1.20 mmol/l (SD 0.69)] in proportion to hyperglycemia [BBGR 0.09 (SD 0.06)]. Breath glucose was elevated more than expected for blood glucose in cystic fibrosis patients [breath 2.04 mmol/l (SD 1.14), BBGR 0.29 (SD 0.17)] and in cystic fibrosis-related diabetes [breath 4.00 mmol/l (SD 2.07), BBGR 0.54 (0.28); P < 0.0001]. These data indicate that 1) this method makes a biologically plausible estimate of respiratory fluid glucose concentration, 2) respiratory fluid glucose concentrations are elevated by hyperglycemia and lung disease, and 3) effects of hyperglycemia and lung disease can be distinguished using the BBGR. This method will support future in vivo investigation of the cause and effect of elevated respiratory fluid glucose in human lung disease.