Background: Data on longitudinal respiratory follow-up after preterm birth in the surfactant era are scarce and of increasing importance, with concerns that preterm survivors are destined for early onset chronic obstructive airway disease. We aimed to comprehensively assess lung function longitudinally from early childhood to mid-childhood in very preterm children (≤32 weeks gestation), and to explore factors negatively impacting on lung function trajectories. Methods: Preterm children (with and without bronchopulmonary dysplasia) and healthy term children as controls were studied. All preterm participants were born at 32 weeks' gestation or earlier at King Edward Memorial Hospital, Perth, WA, Australia, between 1997 and 2003. Bronchopulmonary dysplasia was defined as at least 28 days of supplemental oxygen requirement as assessed at 36 weeks' post-menstrual age. Spirometry, oscillatory mechanics, gas exchange, lung volumes, and respiratory symptoms were assessed at three visits, two in early childhood (4–8 years) and one in mid-childhood (9–12 years). CT of the chest was done in preterm children in mid-childhood. Respiratory symptoms were documented via questionnaire at each visit. Data were analysed longitudinally using linear mixed models. Findings: 200 very preterm children (126 with bronchopulmonary dysplasia and 74 without bronchopulmonary dysplasia) and 67 healthy term control children attended 458 visits between age 4 and 12 years. Chest CT was done on 133 preterm children at a mean age of 10·9 (SD 0·6) years. Preterm children, with and without bronchopulmonary dysplasia, had declines in spirometry z-scores over time compared with controls: forced expiratory volume in 1 s (FEV1), forced expiratory flow at 25–75% of the pulmonary volume, and FEV1/forced vital capacity all declined by at least 0·1 z-score per year in children with bronchopulmonary dysplasia (all p<0·001). Respiratory mechanics and gas exchange also deteriorated over time in children with bronchopulmonary dysplasia (relative to term controls, respiratory system reactance at 8 Hz decreased by −0·05 z-score per year [95% CI −0·08 to −0·01; p=0·006] and diffusing capacity of the lungs for carbon monoxide decreased by −0·03 z-score per year [95% CI −0·06 to −0·01; p=0·048]). Preterm children with bronchial wall thickening on chest CT (suggestive of inflammation) had bigger decreases in spirometry outcomes through childhood. For example, children with bronchial wall thickening on chest CT had an FEV1 z-score decline of −0·61 (95% CI −1·03 to–0·19; p=0·005) more than those without. Similarly, children exposed to tobacco smoke, those with earlier gestation, or those requiring more neonatal supplemental oxygen declined at a faster rate. Interpretation: Lung function trajectories are impaired in survivors of very preterm birth. Survivors with bronchopulmonary dysplasia, ongoing respiratory symptoms, or CT changes reflecting inflammation have the poorest trajectories and might be at increased risk of lung disease in later life. Close targeted pulmonary follow-up of these individuals is necessary. Funding: National Health and Medical Research Council grants APP634519, APP1073301 (to SJS), APP1077691 (to JJP), and APP1025550 (to GLH), Princess Margret Hospital Foundation, and Raine Medical Foundation.