We study techniques for identifying highly boosted top jets, where the subsequent top decay products are not isolated. For hadronic boosted tops, we consider variables which probe the jet substructure in order to reduce the background from QCD jets with large invariant mass. Substructure variables related to two-body kinematics are least sensitive to the modeling of parton shower, while those which involve multi-body kinematics may still have discrimination power. For leptonic boosted tops, we consider variables which characterize the separation between the lepton—although not isolated by conventional criterion—and the hadronic activity in the top jet. Such variables are useful in reducing the backgrounds both from heavy-flavor jets and from accidental jet-lepton overlap. We give numerical estimates of the top identification efficiency versus background rejection rate as a functions of cuts on these variables, and find that these variables offer additional useful information above invariant mass alone.