Background: The target coverage for radiotherapy of early-stage lung cancer was evaluated using two different CT techniques. Material and methods: A conventional planning CT scan and two limited scans of the tumor region were performed in seven patients with peripheral tumors. Three 'slow' scans (slice thickness 4 mm, index 3 mm, revolution time 4 s/slice) were then performed, followed by three-dimensional image registration. Planning target volumes (PTV) were generated using these GTV-PTV margins: (a) 1 cm (PTV1.0); (b) 1.5 cm (PTV1.5); and (c) 0.9, 1.0, and 0.9 cm ('PTVclinical') when set-up errors are avoided. Results: PTVs derived from three 'slow' scans missed 1.9% of the volume derived from three planning scans for an immobile tumor and 9.3% in the case of a mobile tumor. For an immobile tumor, PTV1.5 achieved comparable coverage to that achieved using PTVclinical, which was generated from three 'slow' scans and a planning scan. For a mobile tumor, PTV1.5 covered only 89% of the volume captured by PTVclinical. PTV1.0 resulted in inadequate target coverage in all the patients. Reductions in potential lung toxicity (V20) were achievable in six patients despite the larger GTVclinical when treatment set-up errors were minimized. Conclusions: PTVs derived using 'slow' CT scans consistently produce superior target coverage than that using conventional scans. This may account for the poor local control observed in stage I lung cancer.