For biped pole-climbing robots (BiPCRs), footholds, indicating a sequence of discrete gripping points from the start to the goal, are necessary for travelling in trussed environments. Graspable region in each climbing cycle is the valuable priori knowledge for consideration of selecting specific grasping point. However, how to evaluate all the grasps within the graspable region so as to provide a reasonable grasp selection strategy, is an interesting and pending issue. In this paper, we present three criteria to evaluate grasps considering the characteristics of the climbing motion of BiPCRs. Based on these criteria, we also propose two strategies to optimally select a grasp from the graspable region. An algorithm is presented for BiPCRs to perform foothold planning in trusses. Simulations are carried out to verify the effectiveness of the criteria, the selection strategies and the foothold planning algorithm.