We have investigated a staphylococcal surface display system for its potential future use as a protein library display system in combinatorial biochemistry. Efficient affinity-based selections require a system capable of fine affinity discrimination of closely related binders to minimize the loss of potentially improved variants. In this study, a significant breakthrough was achieved to avoid biases due to potential cell-to-cell variations in surface expression levels, since it was found that a generic protein tag, present within the displayed recombinant surface proteins on the cells, could be successfully employed to obtain normalization of the target-binding signal. Four mutated variants of a staphylococcal protein A domain with different affinity to human IgG were successfully expressed on the surface of recombinant Staphylococcus carnosus cells. The system was evaluated for affinity-based cell sorting experiments, where cell-displayed protein A domains with an 8-fold difference in target affinity were mixed at a ratio of 1:1000 and sorted using FACS. Enrichment factors around 140-fold were obtained from a single round of sorting under normal library sorting conditions when the top 0.1% fraction having the highest antigen binding to surface expression level ratio was sorted. The results demonstrate that the system would have a potential as a selection system in protein library display applications, and the normalization strategy should indeed make it possible to achieve fine affinity discriminations in future library selections.