A hydrocarbon trapping system for cold start emissions was constructed and tested using two types of carbonaceous adsorbents provided by Corning, Inc. One was made by combining activated carbon with an organic binder and extruding it into a honeycomb, and the other by depositing a carbon coating on a ceramic monolith. The tests were carried out on an engine in a dynamometer laboratory to characterize the performance of the carbon elements under transient cold start conditions. Performance was evaluated by continuously measuring exhaust gas hydrocarbon concentrations upstream and downstream of the trap, using conventional emissions consoles. Samples were also collected for off-line analysis of individual hydrocarbon species using gas chromatography to examine differences in adsorption of individual species. The speciated hydrocarbon data were used to distinguish between the mass trapping efficiency and a reactivity-based trapping efficiency of the adsorbant traps. Trap performance was investigated using both a 91 RON base gasoline and a California reformulated fuel. The two carbon adsorbents were found to have similar abilities to collect total hydrocarbons, with the carbon coated monolith achieving somewhat higher effiencies, presumably due to its greater geometric surface area. The traps differed in their ability to release adsorbed hydrocarbons when treated with a heated air purge. The tolerance to higher temperature of the carbon coated monolith sample could be exploited for more efficient purging of the trap using a stream of heated air. Speciation of samples collected before and after the traps revealed that high molecular weight hydrocarbons were trapped very efficiently, while low molecular weight species were not. Because atmospheric reactivity varies among hydrocarbons, reactivity trapping efficiency differed from mass trapping efficiency. The results obtained illustrate the effect that fuel selection has when determining the mass- and reactivity-based trapping efficiency of cold trap adsorbants.