An experimental and numerical study was carried out to simulate the diesel spray behavior during cold starting conditions inside two single-cylinder optically accessible engines. One is an AVL single-cylinder research diesel engine converted for optical access; the other is a TACOM/LABECO engine retrofitted with mirror-coupled endoscope access. The first engine is suitable for sophisticated optical diagnostics but is constrained to limited consecutive fuel injections or firings. The second one is located inside a micro-processor controlled cold room; therefore it can be operated under a wide range of practical engine conditions and is ideal for cycle-to-cycle variation study.The intake and blow-by flow rates are carefully measured in order to clearly define the operation condition. In addition to cylinder pressure measurement, the experiment used 16-mm high-speed movie photography to directly visualize the global structures of the sprays and ignition process. A photodiode was used to quantify the ignition and combustion event through the endoscope for the LABECO engine. Limited tests using two laser diagnostics techniques were also carried out on the AVL engine to help characterize the key physical processes involved in the fuel-air mixing. Planar laser-induced exciplex fluorescence was used to visualize the vaporizing diesel sprays; phase Doppler analyzer system was also employed to perform the simultaneous measurements of droplet size and velocity. A modified version of KIVA-II was used to evaluate the blowby, spray and multistep preignition kinetics models for the AVL engine data.The results showed that optically accessible engines provide very useful information for studying the diesel cold starting conditions, which also provide a very critical test for diesel combustion models. The pre-ignition chemistry showed great sensitivity to the compressed air temperatures and the fuel-air mixting processes. KIVA with a modified shell model responds accordingly to the change of inlet air temperatures and fuel injection parameters. However, other submodels do not have enough sensitivity to simulate the starting of diesel engine without careful validation and further improvements. Blowby model is very critical for cold starting condition but current model underestimates the temperature effect. Widely used empirical correlations and breakup models tend to overpredict the penetration of the low pressure diesel spray typically observed during cold starting.