This experimental study focuses on the effects of the reactivity separation between the port injected fuel and the direct injection fuel, the amount of external-cooled exhaust gas recirculation (EGR), and the direct injection timing of the high reactivity fuel on Reactivity Controlled Compression Ignition (RCCI) combustion. The experiments were conducted on a light-duty, single-cylinder diesel engine with a production GM/Isuzu engine head and piston and a retrofitted port fuel injection system. The global charge-mass equivalence ratio, ϕ′, was fixed at 0.32 throughout all of the experiments. To investigate the effects of the fuel reactivity separation, different Primary Reference Fuels (PRF) were port injected, with the PRF number varying from 50 to 90. To investigate the effects of EGR, an EGR range of 0 to 55% was used. To investigate the effects of the injection timing, an injection timing window of −65 to −45 degrees ATDC was chosen.The results indicate that there are several tradeoffs. First, decreasing the port injected fuel reactivity (increasing the PRF number) delays combustion phasing, decreases the combustion efficiency by up to 9%, increases the gross indicated thermal efficiency up to 22%, enhances the combustion sensitivity to the direct injection timing, and slightly increases the UHC, CO, and NOx emissions. Second, increasing the EGR percentage delays combustion phasing, lowers the peak heat release rate, and lowers the NOx emissions. The combustion efficiency first increases and then decreases with EGR percentage for high reactivity fuels (low PRF number), but only decreases for low reactivity fuels. Finally, delaying the injection timing advances combustion phasing and increases the combustion efficiency, but decreases the gross indicated thermal efficiency and increases the NOx emissions. Across all of the experiments, delays in CA50 increase the gross indicated thermal efficiency and decrease the combustion efficiency, which represents an inherent tradeoff for RCCI combustion on a light-duty engine.