Test results are affected by temporal and spatial variability, process fluctuations, sampling team and laboratory vagaries and techniques, as well as the inherent characteristics of the test method. These errors all affect measurements generated during multi-run compliance test results. Overall measurement variability is made up of process variability which describes the expected emissions range after removing the effect of measurement method uncertainty. Measurement uncertainty is substantially larger than generally assumed, but can be accommodated when determining compliance with regulatory limits and contractual guarantees. (C) 2000 Elsevier Science B.V. All rights reserved.