Summary: A hallmark of adaptive immune responses is the generation of long-lived protection after primary exposure to a pathogen. In humoral responses, this protection stems from a combination of sustained antibody titers and long-lived memory B cells (MBCs), with the former deriving from long-lived plasma cells (PCs). Both types of cell are thought to primarily derive from the germinal center (GC), a unique structure that forms during the immune response to many types of antigenic stimuli. GCs are seeded by antigen-specific B and T cells that were previously activated in the early stages of the response. The GC does not directly or immediately generate effector function; rather, it is a site of intense B-cell proliferation and cell death. GC B cells undergo both somatic hypermutation and isotype switch, and a Darwinian process very efficiently selects B cells with higher fitness for survival and expansion. GC B cells adopt a unique activation and transcriptional state, and the cells become poised to differentiate to either MBCs or PCs. Despite this general understanding of the events in the GC, the mechanisms that control both affinity selection as well as differentiation have not been well worked out. In this review, we address what is known about what determines whether GC B cells become MBCs or PCs. This is discussed in the broader context of the origins of both cell types, whether from the GC or potentially other sources. We present a model encompassing recent data from several laboratories including our own that suggests that the GC undergoes a temporal switch that alters the nature of its output from MBCs to PCs as the response progresses. We will discuss B-cell receptor signaling in the GC as it relates to potential mechanisms for affinity-based selection during the reaction.