The use of ammonia (NH3), a low life-cycle carbon fuel, is an increasingly popular pathway towards decarbonization in the marine and other sectors. However, NH3 possesses low reactivity and flame speed, making its use in internal combustion engines challenging. Additionally, combustion of NH3 can produce incomplete combustion, combustion instability, and toxicity concerns related to fuel slip. Therefore, robustly igniting the fuel and promoting effective flame propagation is critical for NH3 usage in engines. In the present study, investigations of NH3 combustion in a 0.4-liter single-cylinder spark-ignited (SI) research engine are carried out experimentally over a range of operating conditions. 100% NH3 operation successfully covers 60% of the speed-load map, while other areas require aid from a secondary fuel. Compared to the gasoline baseline, 7 percentage points higher peak efficiency is realized by NH3, and nitrogen oxides (NOx) emissions are reduced by two thirds. Separately, computational fluid dynamics (CFD) investigations are used to understand the cyclic variability associated with NH3 SI combustion and passive pre-chamber combustion. The observed cyclic variability in Reynolds Averaged Navier-Stokes (RANS) CFD is introduced by the variability of turbulent kinetic energy (TKE) distribution and flow fields, combined with NH3’s less reactive flame chemistry that amplifies the sensitivity to this turbulence. It is concluded that the cyclic variability observed with RANS CFD is an effective measure of combustion robustness concerning fuel and flame sensitivity. The passive pre-chamber shows promising simulation results compared to SI, resulting in higher thermal and combustion efficiency and reduced combustion instability. These results illustrate the challenges associated with developing and simulating single-fuel NH3 combustion engines and indicate promising routes forward.