A huge portion of solar photovoltaic (PV) systems is installed in residential sector, recently. Considering the uncertain nature of PV generation, battery units are coupled with these systems to provide a better performance when trading power with upstream network. Although, pairing PVs with battery units (PV-battery) provides a promising operation, the uncertainties of load and solar radiation poses a noticeable effect on optimal sizing of such systems. Therefore, an appropriate characterization of these uncertainties is essential when sizing residential PV-battery systems. This can assist decision makers to 1) optimally design their system from both economical and operational perspectives, and 2) avoid non-optimal sizing decisions, regarding the huge investment costs of these systems. This study presents a robust optimization (RO) approach for sizing residential PV-battery systems, characterizing the uncertainties of load and solar radiation. The proposed model determines the optimal capacity of a residential PV-battery system while minimizing its operation costs considering the uncertain parameters' variations. The RO model is developed as a max-min optimization problem. According to the obtained results, the proposed robust model gives a greater sizing solution (50% increase in PV capacity), which is more reliable when the uncertain parameters deviate from the forecasted values.