The exhaust filtration analysis system (EFA) developed at the University of Wisconsin–Madison was used to perform microscale filtration experiments on cordierite filter samples using particulate matter (PM) generated by a spark ignition direct injection (SIDI) engine fueled with gasoline. A scanning mobility particle sizer (SMPS) was used to characterize running conditions with four distinct particle size distributions (PSDs). The distributions selected differed in the relative number of accumulation versus nucleation mode particles. The SMPS and an engine exhaust particle sizer (EEPS) were used to simultaneously measure the PSD downstream of the EFA and the real-time particulate emissions from the SIDI engine to determine the evolution of filtration efficiency (FE) during filter loading. Cordierite filter samples with properties representative of diesel particulate filters (DPFs) were loaded with PM from the different engine operating conditions. The results were compared to understand the impact of PSD on filtration performance as well as the role of accumulation mode particles on the diffusion capture of PM. The most penetrating particle size (MPPS) was observed to decrease as a result of particle deposition within the filter substrate. In the absence of a soot cake, the penetration of particles smaller than 70 nm was seen to gradually increase with time, potentially due to increased velocities in the filter as flow area reduces during filter loading, or due to decreasing wall area for capture of particles by diffusion. Particle re-entrainment was not observed for any of the operating conditions.