Purpose: Fluoroscopy is widely used for evaluating tumor mobility in radiotherapy planning. Lung tumor mobility was scored using virtual fluoroscopy, and this was compared to mobility derived from contoured tumors in all phases of a respiration-correlated (or 4D) CT scan. Methods and materials: 4DCT datasets were reviewed and 29 patients were identified in whom tumors were visible on anterior-posterior fluoroscopy views. Mobility in all directions was estimated on fluoroscopy movie loops by four clinicians. These results were compared to mobility measured from contoured tumor volumes in all phases of the same 4DCT. Internal target volumes (ITV) were generated for both approaches. Results: In eight patients, fluoroscopy did not allow for tumor mobility to be assessed in at least one direction. No significant inter-clinician variation was observed with respect to fluoroscopic assessment of mobility. Clinicians systematically overestimated mobility in all three directions (p < 0.05). The mean ITVs derived using fluoroscopy were 52.2% larger than those derived using 4DCT contours, but the individual ITVs were smaller in three patients. Conclusion: Use of virtual fluoroscopy generally overestimates the mobility of visible lung tumors, and results in irradiation of unnecessarily large target volumes. In contrast, use of 4DCT minimizes the risk of normal tissue toxicity.