To improve fuel economy and reduce regulated emissions spark-ignition engines are equipped with a large number of control actuators, motivating the use of model-based ignition timing prediction strategies. Model-based ignition timing strategies require a target combustion phasing for proper calibration, generally defined by the crank angle location where fifty percent of the air/fuel mixture is burned (CA50). When fuel type is altered the target CA50 must be updated in the ‘knock region’ to avoid engine damage while maintaining the highest possible efficiency. This process is particularly important when switching between gasoline and E85 because they have vastly different octane ratings.A semi-physical virtual octane sensor, based on an Arrhenius function combined with a quasi-dimensional turbulent flame entrainment combustion model, is described that identifies the size of the knock region for a given fuel. The combustion duration model is used to calculate cylinder pressure and temperature which are analyzed with an Arrhenius knock prediction model that accounts for the negative temperature coefficient and air/fuel ratio. An algorithm is developed to identify the “best achievable” combustion phasing and update the target desired combustion phasing accordingly. The algorithm operates off-line once the fuel octane number is observed to have changed, and then revised combustion phasing targets are calculated throughout the knock region. Experimental measurements and simulations are used to correlate and validate the algorithm with both gasoline and E85. Results are presented and conclusions are drawn.