Methane (CH4) emissions from animals represent a significant contribution to anthropogenically produced radiatively active trace gases. Global and national CH4 budgets currently use predictive models based on emission data from laboratory experiments to estimate the magnitude of the animal source. This paper presents a method for measuring CH4 from animals under undisturbed field conditions and examines the performance of common models used to simulate field conditions. A micrometeorological mass difference technique was developed to measure CH4 production by cattle in pasture and feedlot conditions. Measurements were made continuously under field conditions, semiautomatically for several days, and the technique was virtually nonintrusive. The method permits a relatively large number of cattle to be sampled. Limitations include light winds (less than approximately 2 m/s), rapid wind direction changes, and high-precision CH4 gas concentration measurement. Methane production showed a marked periodicity, with greater emissions during periods of rumination as opposed to grazing. When the cattle were grazed on pasture, they produced .23 kg CH4 x animal(-1) x d(-1), which corresponded to the conversion of 7.7 to 8.4% of gross energy into CH4. When the same cattle were fed a highly digestible, high-grain diet, they produced .07 kg CH4 x animal(-1) x d(-1), corresponding to a conversion of only 1.9 to 2.2% of the feed energy to CH4. These measurements clearly document higher CH4 production (about four times) for cattle receiving low-quality, high-fiber diets than for cattle fed high-grain diets. The mass difference method provides a useful tool for "undisturbed" measurements on the influence of feedstuffs and nutritional management practices on CH4 production from animals and for developing improved management practice for enhanced environmental quality.