Efficiency of spark ignition (SI) engines is limited towards high loads by the occurrence of knock, which is linked to the octane number of the fuel. Running the engine at its optimal efficiency requires a high octane number at high load whereas a low octane number can be used at low load.Current project aims at developing an “Octane on Demand” (OOD) concept: the fuel octane number is adjusted “on demand” to prevent knock occurrence by adapting the fuel RON injected in the combustion chamber. Thus, the engine cycle efficiency is increased by always keeping combustion phasing at optimum. This is achieved by a dual fuel injection strategy, involving a low-RON base-fuel and a high-RON octane booster. The ratio of fuel quantity on each injector is adapted to fit the RON requirement function of engine operating conditions.This OOD concept requires a good characterization of the octane requirement needed to run the engine at its optimal efficiency over the entire map. It also involves choosing the best dual fuel combination, including a base-fuel and an octane booster to fit this concept.To find the best fuel combination, three different octane boosters (ethanol, reformate and a blend of butanol isomers (SuperButolTM)) and two base-fuels (a very low-RON naphtha (71 RON) and a Non-Oxygenated Gasoline (NOG, 91 RON)) are tested on a test bench using an up-to-date gasoline direct plus indirect injection multicylinder engine. The way the two fuels are injected (GDI or PFI) is also investigated. Results reveal that the injection configuration has quite a low effect on the octane booster demand (to keep the engine at its optimal combustion phasing). As shown in previous work, ethanol is confirmed to be a remarkable octane booster, for its favorable sensitivity S (defined as S = RON - MON). It is also shown that the base-fuel, compared to the high octane fuel, has a weaker effect on the octane booster requirement, promoting the use of a less processed naphtha-based low-RON fuel (71) as a base-fuel.According to 0D-simulations, about 4% CO2 savings are expected over the WLTP cycle, considering a light C sedan car powered with an up-to-date 1.6L. The more loaded the driving cycle, the larger the CO2 gains: up to 25% CO2 savings are obtained in full load conditions. Finally, work is in progress to build a complete dual fuel democar.