BACKGROUND: Low-dose spiral CT of the chest effectively detects early-stage lung cancer in high-risk individuals. The high rate of benign nodules and issues of making a differential diagnosis are critical factors that currently hamper introduction of large-scale screening programmes. We investigated the efficacy of repeated yearly spiral CT and selective use of positron emission tomography (PET) in a large cohort of high-risk volunteers. METHODS: We enrolled 1035 individuals aged 50 years or older who had smoked for 20 pack-years or more. All patients underwent annual low-dose CT, with or without PET, for 5 years. Lesions up to 5 mm were deemed non-suspicious and low-dose CT was repeated after 12 months (year 2). FINDINGS: By year 2, 22 cases of lung cancer had been diagnosed (11 at baseline, 11 at year 2). 440 lung lesions were identified in 298 (29%) participants, and 95 were recalled for high-resolution contrast CT. PET scans were positive in 18 of 20 of the identified cancer cases. Six patients underwent surgical biopsy for benign disease because of false-positive results (6% of recalls, 22% of invasive procedures). Complete resection was achieved in 21 (95%) lung cancers, 17 (77%) were pathological stage I (100% at year 2), and the mean tumour size was 18 mm. There were no interval lung cancers in the 2.5 years of follow-up (average time on study from randomisation to last contact), although 19 individuals were diagnosed with another form of cancer (two deaths and 17 non-fatal admissions). INTERPRETATION: Combined use of low-dose spiral CT and selective PET effectively detects early lung cancer. Lesions up to 5 mm can be checked again at 12 months without major risks of progression.