Natural Gas (NG) is an alternative fuel which has attracted a lot of attention recently, in particular in the US due to shale gas availability. The higher hydrogen-to-carbon (H/C) ratio, compared to gasoline, allows for decreasing carbon dioxide emissions throughout the entire engine map. Furthermore, the high knock resistance of NG allows increasing the efficiency at high engine loads compared to fuels with lower knock resistance. NG direct injection (DI) allows for fuel to be added after intake valve closing (IVC) resulting in an increase in power density compared to an injection before IVC. Steady-state engine tests were performed on a single-cylinder research engine equipped with gasoline (E10) port-fuel injection (PFI) and NG DI to allow for in-cylinder blending of both fuels. Knock investigations were performed at two discrete compression ratios (CR), 10.5 and 12.5. Operating conditions span mid-load, wide-open-throttle and boosted conditions, depending on the knock response of the fuel blend. Blended operation was performed using E10 gasoline and NG. An additional gasoline type fuel (E85) with higher knock resistance than E10 was used as a high-octane reference fuel, since the octane rating of E10-NG fuel blends is unknown. Spark timing was varied at different loads under stoichiometric conditions in order to study the knock response as well as the effects on performance and efficiency. As anticipated, results suggest that the knock resistance can be increased significantly by increasing the NG amount. Comparing the engine operation with the least knock resistant fuel, E10 PFI, and the fuel blend with the highest knock resistance, 75% NG DI, shows an increase in indicated mean effective pressure of about 9 bar at CR 12.5. The usage of reference fuels with known knock characteristics allowed an assessment of knock characteristic of intermediate E10-NG blend levels. Mathematical correlations were developed allowing characterizing the occurrence of knocking combustion by using the Livengood-Wu knock integral. For most of the fueling strategies and operating conditions, the mathematical correlations show good agreement when compared to experimental data.