Halo stars with unusually high radial velocity ( ” hypervelocity ” stars , or HVS ) are thought to be stars unbound to the Milky Way that originate from the gravitational interaction of stellar systems with the supermassive black hole at the Galactic center . We examine the latest HVS compilation and find peculiarities that are unexpected in this black hole-ejection scenario . For example , a large fraction of HVS cluster around the constellation of Leo and share a common travel time of \sim 100 - 200 Myr . Furthermore , their velocities are not really extreme if , as suggested by recent galaxy formation models , the Milky Way is embedded within a 2.5 \times 10 ^ { 12 } h ^ { -1 } M _ { \odot } dark halo with virial velocity of \sim 220 km/s . In this case , the escape velocity at \sim 50 kpc would be \sim 600 km/s and very few HVS would be truly unbound . We use numerical simulations to show that disrupting dwarf galaxies may contribute halo stars with velocities up to and sometimes exceeding the nominal escape speed of the system . These stars are arranged in a thinly-collimated outgoing “ tidal tail ” stripped from the dwarf during its latest pericentric passage . We speculate that some HVS may therefore be tidal debris from a dwarf recently disrupted near the center of the Galaxy . In this interpretation , the angular clustering of HVS results because from our perspective the tail is seen nearly “ end on ” , whereas the common travel time simply reflects the fact that these stars were stripped simultaneously from the dwarf during a single pericentric passage . This proposal is eminently falsifiable , since it makes a number of predictions that are distinct from the black-hole ejection mechanism and that should be testable with improved HVS datasets .