Summary Talk on Quark-Hadron Duality

I ascribe the origin of Bloom-Gilman duality in DIS to a separation of scales between the hard subprocess and soft resonance formation. The success of duality indicates that the subprocesses of exclusive form factors are the same as in DIS. The observed dominance of the longitudinal structure function at large x in \pi N \to \mu^+\mu^- X can explain why local duality works for DIS with a pion target. The failure of duality in semi-exclusive processes indicates that high momentum transfer t is not sufficient to make the corresponding subprocesses compact.