Background: Indeterminate non-calcified lung nodules are frequent when low-dose spiral computed tomography (LD-CT) is used for lung cancer screening. We assessed the diagnostic utility of a non-invasive work-up protocol for nodules detected at baseline in volunteers enrolled in our single-centre screening trial, and followed for at least 1 year. Methods: 5201 high-risk volunteers, recruited over 1 year from October 2004, underwent baseline LD-CT; 4821 (93%) returned for the first repeat LD-CT. Nodules ≤5 mm underwent repeat LD-CT at 1 year; nodules 5.1-8 mm underwent LD-CT 3 months later; lesions >8 mm received combined CT-positron emission tomography (CT-PET). A subset of nodules >8 mm was studied by CT with contrast. Protocol failures were delayed diagnosis with disease progression beyond stage I, and negative surgical biopsy. Results: 2754 (53%) volunteers presented one or more non-calcified nodules. Ninety-two lung cancers were diagnosed: 55 at baseline and 37 at annual screening (66% stage I). Among the 37 incident cancers, 17 had a baseline nodule that remained stage I, 7 had a baseline nodule that progressed beyond stage I, and 13 presented a new malignant nodule. Baseline and annual cancers were 79 (1.5%) and 13 (0.2%), respectively. In 15 of 104 (14%) invasive diagnostic procedures, the lesion was benign. Sensitivity, and specificity were 91 and 99.7%, respectively, for the entire protocol; 88 and 93% for CT-PET; and 100 and 59% for CT with contrast. Conclusions: The protocol limits invasive diagnostic procedures while few patients have diagnosis delay, supporting the feasibility of lung cancer screening in high-risk subjects by LD-CT. Nevertheless further optimization of the clinical management of screening-detected nodules is necessary. © 2008 Elsevier Ireland Ltd. All rights reserved.