Engine-out, cold-start (from 20°C) UHC emissions from a contemporary 2.0 4-cylinder engine with swirl control were measured with FID and FT-IR. The steady-state, end of test operation was 1500 rpm, 2.6 bar BMEP (25% load) and stoichiometric mixture. Four fuel systems were employed pintle-type port-injected gasoline, air-forced port-injected gasoline, port-injected propane, and premixed propane. These fuel systems were chosen to separate effects of fuel atomization, vaporization, and fuel-air mixing. Each system was optimized with respect to injector targeting, injection timing, mixture enrichment, and spark advance. Open-valve injection timing increased UHC emissions more with the pintle-type injector than with the air-forced, system. UHC emissions with propane injection were minimized with open valve injection.With gasoline, the air-forced (SMD < 10 μm) injector reduced, by 25% the total UHC emitted during the first two minutes of operation as compared to the pintle-type (SMD ≅ 150 μm) injector and in terms of brake specific hydrocarbon emissions (BSHC). In terms of brake specific ozone forming potential (BSOFP), the reduction was 13%. With propane, additional reductions of 40% BSHC and 32% BSOFP were obtained. With all fuel systems, the additional UHC emissions during cold-start consisted of unburned fuel components and occurred mostly during the first 10 seconds, when the engine load had not yet stabilized. The main, difference between the UHC mass emissions from gasoline and propane was the absence of aromatics for the latter.It is argued that the causes of cold-start UHC emissions are primarily crevices and cold walls for premixed propane, to which mostly cycle-to-cycle injection variation must be added for injected propane. To those mechanisms, we must add oil layer and deposit effects for the air-forced injector, which acted as net sinks, rather than sources, in our case because clean oil was used. Finally, to all of the above sources, the required enrichment and the coarse spray must be added for the commonly used pintle-type injector.The most important overall conclusion is that cycle-resolved control of the equivalence ratio may be the most effective way of reducing cold-start UHC emissions.