Knowledge of the emission rate and release characteristics is necessary for estimating pollutant fate and transport. Because emission measurements at a facility's fence line are generally not readily available, environmental agencies in many countries are using emission factors (EFs) to indicate the quantity of certain pollutants released into the atmosphere from operations such as welding. The amount of fumes and metals generated from a welding process is dependent on many parameters, such as electrode composition, voltage, and current. Because test reports on fume generation provide different levels of detail, a common approach was used to give a test report a quality rating on the basis of several highly subjective criteria; however, weighted average EFs generated in this way are not meant to reflect data precision or to be used for a refined risk analysis. The 95% upper confidence limit (UCL) of the unknown population mean was used in this study to account for the uncertainty in the EF test data. Several parametric UCLs were computed and compared for multiple welding EFs associated with several mild, stainless, and alloy steels. Also, several nonparametric statistical methods, including several bootstrap procedures, were used to compute 95% UCLs. For the nonparametric methods, a distribution for calculating the mean, standard deviation, and other statistical parameters for a dataset does not need to be assumed. There were instances when the sample size was small and instances when EFs for an electrode/process combination were not found. Those two points are addressed in this paper. Finally, this paper is an attempt to deal with the uncertainty in the value of a mean EF for an electrode/process combination that is based on test data from several laboratories. Welding EFs developed with a defined level of confidence may be used as input parameters for risk assessment.