In　the　public　transport　path　plan,　the　goal　is　to　generate　a　path　using　buses　or　subway　in　the　smallest　expected　time.　Because　bus　arrival　time　is　not　determined　but　instead　may　be　random,　in　this　paper,　we　studied　the　constrained　reliable　shortest　path　search　problem　that　arises　in　transit　route　plan　when　waiting　and　in-vehicle　traveling　time　are　uncertain.　The　reliable　path　under　the　duration　constraint　is　searched　heuristically　by　avoiding　the　high　time-variation　links.　An　in-depth　experimental　performance　analysis　of　the　proposed　methods　on　real　transit　graphs　shows　the　efficiency　of　the　algorithm.