Bronchoscopic biopsy of lymph nodes is an important step in staging lung cancer. Lymph nodes, however, lie behind the airway walls and are near large vascular structures - all of these structures are hidden from the bronchoscope's field of view. Previously, we had presented a computer-based virtual bronchoscopic navigation system that provides reliable guidance for bronchoscopic sampling. While this system offers a major improvement over standard practice, bronchoscopists told us that target localization- lining up the bronchoscope before deploying a needle into the target - can still be challenging. We therefore address target localization in two distinct ways: (1) automatic computation of an optimal diagnostic sampling pose for safe, effective biopsies, and (2) a novel visualization of the target and surrounding major vasculature. The planning determines the final pose for the bronchoscope such that the needle, when extended from the tip, maximizes the tissue extracted. This automatically calculated local pose orientation is conveyed in endoluminal renderings by a 3D arrow. Additional visual cues convey obstacle locations and target depths-of-sample from arbitrary instantaneous viewing orientations. With the system, a physician can freely navigate in the virtual bronchoscopic world perceiving the depth-of-sample and possible obstacle locations at any endoluminal pose, not just one pre-determined optimal pose. We validated the system using mediastinal lymph nodes in eleven patients. The system successfully planned for 20 separate targets in human MDCT scans. In particular, given the patient and bronchoscope constraints, our method found that safe, effective biopsies were feasible in 16 of the 20 targets; the four remaining targets required more aggressive safety margins than a "typical" target. In all cases, planning computation took only a few seconds, while the visualizations updated in real time during bronchoscopic navigation.