According to the prevailing "clock" model, chromosome decondensation and nuclear envelope reassembly during mitotic exit are byproducts of Cdk1 inactivation at the metaphase-anaphase transition, controlled by the spindle assembly checkpoint. However, mitotic exit was recently shown to be a function of chromosome separation during anaphase, assisted by a midzone Aurora B phosphorylation gradient -the "ruler" model. Here we reconciled both models by showing that Cyclin B1 degradation continues during anaphase in Drosophila, mouse and human cells, including primary tissues. This required APC/C Cdh1 activity, and failure to degrade Cyclin B1 during anaphase prevented mitotic exit in a Cdk1-dependent manner. Cyclin B1 localization and half-life during anaphase depended on kinesin-6, which targets Aurora B to the spindle midzone.Mechanistically, we show that anaphase duration is regulated by Aurora Bmediated phosphorylation of Cyclin B1. We propose that a crosstalk between molecular "rulers" and "clocks" licenses mitotic exit only after proper chromosome separation.