The ability to accurately measure ammonia emissions from farm buildings is an important issue both forestablishing emissions regulations and for effective evaluation of mitigation techniques. Most techniques for measuringemissions rely on subsampling within a space. This study examines the influence of subsampling under a variety ofconditions on estimated ammonia recoveries. Tests were made using a large environmental chamber with controlled releasesof ammonia from a gas cylinder from one of two positions within the chamber. Ammonia concentrations were measured incontinuous air samples from either a single exhaust duct or a sampling port in the exhaust plenum where exhaust gases werewell mixed. The chamber temperature was maintained at 22.2. C with an airflow of 10.5 air exchanges per hour. Formeasurements made in the exhaust duct, ammonia recoveries were 217% when the release position and the sampling ductwere aligned in terms of chamber airflow, and were 52% when the positions were not aligned. Placing a wooden barrierbetween aligned release and measurement positions only reduced ammonia recoveries to 173%. In contrast, using anoscillating fan to disperse the ammonia release reduced ammonia recoveries to 78%. When measurements were made in theplenum, recoveries were essentially 100%, and placement of a continuously wetted barrier between the release position andthe exhaust ducts did not influence recoveries. It is suggested that measurement of ammonia emissions be restricted to farmstructures with welldefined airflows and a limited number of exhaust openings, and that the most accurate method forestimating ammonia emission rates would be to collect and mix all of the exhaust streams from a structure prior tocontinuously measuring ammonia concentration and airflow.