Steady state real time optimization (RTO) of a reactor separator recycle process is studied. The reaction A + B -> C occurs in the reactor with the separator recycling unreacted A and B and recovering nearly pure C bottoms product. Two operating scenarios, Mode I (fixed throughput) and Mode II (maximum throughput), are considered. In Mode I, the column boil-up is minimized using available degrees-of-freedom (dofs). In Mode II, the dofs are optimized for maximizing throughput. It is shown that the reactor B composition set point is unconstrained and economically dominant in both Mode I and Mode II. Its optimum value is efficiently obtained by fitting a simple reaction kinetic model to the available plant flow and reactor composition data (current and past) and adjusting the best fit the current temperature profile. Results show that RTO improves economic benefit by up to 10% over constant set point operation. separator tray section efficiencies to best fit the current temperature profile. Results show that RTO improves economic benefit by up to 10% over constant set point operation.