We present a feature-aided approach to multiple-target tracking built upon tree-search tracking algorithms. Using a search tree to represent the target state space, the tracker navigates the tree to identify the most likely sequence of states visited by the target(s). The search for new targets and for new states of existing targets is governed by path metrics that are proportional to the posterior state distribution and incorporate the likelihood of observed feature values. Features may be assumed to follow a given statistical model, or their probability density functions may be estimated empirically using feature history stored with each track in the tree. The performance of the proposed feature-aided tracker is evaluated on the CLUTTER09 dataset. Improvement with respect to target detection/tracking and clutter rejection is evaluated relative to tracking in the absence of feature information.