Abstract The complexity of the optimization problem arises in multistep model predictive control for power electronics as they are discrete by nature and have predefined control actions given as integer control variables. Generally, Sphere Decoding Algorithm (SDA) is used to solve the optimization problem. In this paper, we present an SDA with an Evolutionary Optimization attitude (EO) to simplify the complex exhaustive search that is brought by the long prediction horizon. The presented technique reconstructs a smaller search area from a large search area which decreases the number of candidate solutions. The performance of the optimization algorithm is evaluated through statistical analysis and computation burden.