Indoor　pedestrian　evacuation　processes　could　be　significantly　affected　by　the　presence　of　guides,　that　is,　safety　management　staff;　how　to　assign　guides　properly　remains　a　challenging　task.　This　question　is　deconstructed　into　single-　and　multiple-exit　scenarios　for　analysis　in　this　study.　The　mechanisms　behind　the　evacuation　dynamics　are　explored　via　a　two-layer　guided　pedestrian　evacuation　model　and　the　corresponding　guide　assignment　strategies　are　proposed.　The　upper　layer　model　deals　with　guide　assignment,　where　random　and　uniform　guide　assignment　schemes,　and　a　newly　proposed　distribution-based　guide　assignment　scheme,　are　embedded,　while　the　lower　layer　model　controls　the　movement　of　evacuees　based　on　a　cellular　automata　model.　Results　show　that　there　are　bifurcate　mechanisms　governing　evacuation　dynamics.　Only　in　single-exit　scenarios　could　an　increased　number　of　guides　lead　to　a　＂sharp　decrease-long　steady　tail＂　tendency　in　total　evacuation　time,　whereas　evacuation　efficiency　in　multiple-exit　scenarios　is　consistent　with　exit　use　equilibrium.　The　three　different　guide　assignment　schemes　bring　about　highly　case-specific　performances　depending　on　the　initial　pedestrian　distribution　and　the　number　of　exits.　The　distribution-based　guide　assignment　scheme　is　preferable　in　most　cases,　especially　in　spaces　with　a　highly　biased　pedestrian　distribution　and　multiple　exits.