Solar thermal systems in residential buildings are generally controlled by two-level controllers, which activate solar thermal or at times with low solar radiation auxiliary energy supply into a thermal storage. Simple controllers do not have any information on actual or expected solar radiation. This leads to interference of auxiliary-and solar heat supply, which reduces the share of solar thermal energy fed into the thermal storage. Increasing accuracy of weather forecast data suggests incorporation of this information in the control algorithm. This work analyzes the maximum potential performance enhancement when applying such an intelligent predictive control. Two solar thermal systems with one auxiliary source respectively are designed in TRNSYS-these systems represent the base case. Further, a number of simulations are conducted with minor variations for the plant parameters-this gives generic results for different system configurations. In addition, each system configuration is altered to mimic the behavior of a plant with intelligent predictive control. Comparison of results indicates an improvement potential up to 10% for annual solar fractions and up to 30% for monthly solar fractions. The performance bound with respect to the annual auxiliary energy savings is approximately 8%.