During the past decades, advances in protein engineering have resulted in the development of various in vitro selection techniques (e.g. phage display) to facilitate discovery of new and improved proteins. The methods are based on linkage between genotype and phenotype and are often performed in successive rounds of selection. Since the resulting output depends on the selection pressures used and the applied strategy, parameters in each round must be carefully considered. In addition, studies have reported biases that can cause enrichment of unwanted clones and/or low correlation between abundance in output and affinity. We have recently developed a selection method based on display of protein libraries on Staphylococcus carnosus and isolation of affinity proteins by fluorescence-activated cell sorting. Here, we compared duplicate selections for affinity maturation using equilibrium binding at different target concentrations and kinetic off-rate selection. The results showed that kinetic selection is efficient for isolation of high-affinity binders and that equilibrium selection at subnanomolar concentrations should be avoided. Furthermore, the reproducibility of the selection was high and a clear correlation was observed between enrichment and affinity. This work reports on the reproducibility of bacterial display in combination with FACS and provides insights into selection design to help guide the development of new affinity proteins.