A systematic investigation of sorbent-trap sampling, which is a method that uses paired sorbent traps to measure total vapor phase mercury (Hg), was carried out at two coal-fired power plants. The objective of the study was to evaluate the effects (if any) on data quality when the following aspects of the sorbent trap method are varied: (a) sorbent trap configuration; (b) sampling time; and (c) analytical technique. Also, the performance of a speciating sorbent trap (i.e., a trap capable of separating elemental Hg from oxidized Hg), developed by the Western Kentucky University's Institute for Combustion Science and Environmental Technology (ICSET), was evaluated by direct comparison against the Ontario Hydro (OH) reference method. Flue gas samples were taken using both "regular" and modified sorbent trap measurement systems. The regular sorbent trap systems used a dual-trap, in-stack sampling technique. The modified systems were equipped with either inertial or cyclone probes and used paired, out-of-stack sorbent traps. Both short-term (1.5 h) and long-term (18 h to 10 days) samples were collected. The OH method was run concurrently during the short-term test runs, to provide reference Hg concentrations. At one facility, mercury concentration data from continuous emission monitoring systems were also recorded during the sorbent trap sampling runs. After sampling, the conventional (nonspeciating) sorbent traps were analyzed for Hg, using either a direct combustion method or a wet-chemistry analytical method (i.e., microwave-assisted digestion coupled with cold vapor atomic absorption spectroscopy). The speciating traps were analyzed only by the direct combustion method, All of the sorbent trap data collected in the study were evaluated with respect to relative accuracy, relative deviation of paired traps, and mercury breakthrough. The in-stack and out-of-stack sampling methods produced satisfactory relative accuracy results for both the short-term and long-term testing. For the short-term tests, the in-stack sampling results compared more favorably to the OH method than did the out-of-stack results. The relative deviation between the paired traps was considerably higher for the short-term out-of-stack tests than for the long-term tests. A one-way analysis of variance (ANOVA), showed a statistically significant difference (p < 0.1) between the direct combustion and wet-chemistry analytical methods used in the study; the results from the direct combustion method were consistently higher than the wet-chemistry results. The evaluation of the speciating mercury sorbent trap demonstrated that the trap is capable of providing reasonably accurate total mercury concentrations and speciation data that are somewhat comparable to data obtained with the OH method. Although the results of the study were informative and promising, further evaluation of both the out-of-stack sampling methods and the speciating sorbent trap is recommended.