A turbocharger turbine is exposed to pulsating flow conditions when it is connected to an engine exhaust system due to the opening and closing of the exhaust valves. However, many radial turbines are designed and tested under steady-state conditions without taking into account these unsteady exhaust flows. In order to seek the optimal aerodynamic design of a radial flow turbine (RFT) under pulsating flow conditions, the present research utilizes a numerical simulation approach to optimize the blade shape of a small-scale mixed flow turbine (MFT) under 50 Hz pulses. This corresponds to a four-stroke, three-cylinder engine rotating at 2000 rpm. In order to understand how a less computationally intensive, steady-state optimization compares, the blade shape was also optimized using the peak power point of the pulse. Three turbine features were modified during the optimization process, including blade cone angle, blade axial location, and blade camber angles. The optimization was carried out using a computational fluid dynamics (CFD)–genetic algorithm (GA) coupled approach, targeting at maximizing both energy-weighted efficiency and energy output during a predefined pulse period. To ensure that the new design maintains a similar matching to the engine, the maximum deviation of turbine swallowing capacity is controlled to within of the baseline for all new blade designs. The design that achieves the maximum pulse cycle-averaged efficiency was produced from unsteady optimization, with a performance benefit of 0.66%. The unsteady optimization also produced a blade shape that delivers the maximum energy output, with an improvement of 5.42%.