During postnatal development of the cerebellum, the number of climbing fibers that innervate individual Purkinje cells decreases from many to one. This is one of the most characterized models of activity-dependent refinement of synaptic circuitry in the mammalian brain. As surplus climbing fibers are eliminated, subcellular location of climbing fiber terminals moves from the soma to the dendrites of Purkinje cells. The role of this dendritic translocation in climbing fiber elimination has been under debate for a long time, but recent studies have significantly changed the view. Traditionally, dendritic translocation was considered neither sufficient nor necessary for climbing fiber elimination, but experimental evidence obtained over the past 5 years indicates crucial roles of dendritic translocation for selectively maintaining one fiber while removing all the others. Here, I provide an overview of (i) several key findings that have contributed for developing the traditional and current views of dendritic translocation, (ii) how our novel experimental approach supports the current view, and (iii) a remaining question that is yet to be answered.