It has been known for many years that the affinity of antibodies for antigen increases with time during an immune response. It is now clear that two processes play fundamental roles in this affinity 'maturation' in the mouse - V gene somatic mutation and antigen affinity-based selection. Exactly how these two processes work in concert is not fully understood. In this article Tim Manser argues that models of affinity maturation based on the assumption that somatic mutation, antigen selection and B-cell division are interdependent may not explain the high efficiency of the process, and he suggests an alternative model.