Optimization of antibody affinity is a hallmark of the humoral immune response. It takes place in hundreds of transient microstructures called germinal centers (GCs). Their function and time-dependent behavior are subjects of active investigation. According to a generally accepted notion, their individual kinetics follows the average kinetics of all GCs present in the observed lymphatic tissue. In this review, we challenge this view and point out, with the help of mathematical simulations, that inferring the kinetics of individual GCs from cross-sectional evaluation of GC kinetics is virtually impossible. Thus, the time course of individual GCs is open to conjecture. For instance, one possible interpretation is that GCs exist for a time span considerably shorter than that of the observed average kinetics. We explore the implications of different temporal organizations of GCs in the light of the hypothesis that GC B-cell emigrants recolonize GC niches. This assumption leads to a view where GCs work in parallel but are linked by recirculation of B-cell emigrants. In this view, interleaved global and local competition provide for an implementation of multiple levels of B-cell selection in affinity maturation. The concepts of iteration, all-or-none behavior, and phasic mutation schedule are discussed in the light of this hypothesis.