Desorption canisters are routinely employed to quantify coalbed gas contents in coals. If purging with inert gas or water flooding is not used, entrapment of air with similar to 78.08 vol.% nitrogen (N(2)) in canisters during the loading of coal results in contamination by air and subsequent overestimates of N(2) in desorbed coalbed gas. Pure coalbed gas does not contain any elemental oxygen (O(2)), whereas air contamination originally includes similar to 20.95 vol-% O(2) and has a N(2)/O(2) volume ratio of -3.73. A correction for atmospheric N(2) is often attempted by quantifying O(2) in headspace gas and then proportionally subtracting atmospheric N(2). However, this study shows that O(2) is not a conservative proxy for air contamination in desorption canisters. Time-series of gas chromatographic (GC) compositional data from several desorption experiments using high volatile bituminous coals from the Illinois Basin and a New Zealand subbituminous coal document that atmospheric O(2) was rapidly consumed, especially during the first 24h. After about 2 weeks of desorption, the concentration of O(2) declined to near or below GC detection limits. Irreversible loss of O(2) in desorption canisters is caused by biological, chemical, and physical mechanisms. The use of O(2) as a proxy for air contamination is justified only immediately after loading of desorption canisters, but such rapid measurements preclude meaningful assessment of coalbed gas concentrations. With increasing time and progressive loss of O(2), the use of O(2) content as a proxy for atmospheric N(2) results in overestimates of N(2) in desorbed coalbed gas. The indicated errors for nitrogen often range in hundreds of %. Such large analytical errors have a profound influence on market choices for CBM gas. An erroneously calculated N(2) content in CBM would not meet specifications for most pipeline-quality gas. (C) 2009 Elsevier B.V. All rights reserved.