Although　researchers　have　discussed　several　aspects　of　so　called　browsing，　such　discussions　remained　mere　armchair　ones　and　were　unable　to　prove　whether　their　claims　correctlydescribed　actual　browsing　behavior．　　　This　research　examined　the　actual　behavior　of　the　subjects　in　bookstores　and　libraries　inorder　to　clarify　the　concept　of　‘browsing’．　　　Research　took　place　in　2　bookstores　and　2　libraries，　where　40　subjects　were　observed　fromtheir　entrance　to　their　exit．　As　a　result，　the　following　actions　were　observed：　Checking　theshelves　one　by　one；　re－checking　the　shelves　while　a　different　book　is　already　in　hand；　picking　upthe　same　book　as　one　the　subject　had　just　put　back；　stopping　to　check　the　books　stacked　flat　ondisplay；　returning　the　book　to　shelf　unopened；　turning　pages　unnaturally　fast；　and　openingpages　at　random．　　　In　addition，　it　was　found　that　different　ways　exist　to　perceive　different　types　of　sources，　andthat　the　subjects　judge　the　sources　according　to　certain　standards．　　　In　conclusion，　browsing　was　classified　into　5　types，　and　was　defined　as　follows：　browsing　isa　means　of　selecting　necessary　material　among　many，　according　to　certain　standards，　whileemploying　every　sense　available　in　order　to　satisfy　a　certain　requirement　for　information，　andwhose　target　remains　vague　at　the　outset．