BACKGROUND: Vascular infection with Chlamydia pneumoniae might boost inflammatory responses that play a pivotal part in neointima formation, which is the main cause of restenosis after stenting. Our aim was to investigate whether or not treatment of C pneumoniae infection with antibiotics prevents restenosis after coronary stent placement. METHODS: We enrolled 1010 consecutive patients with successful coronary stenting into a randomised, double-blind trial. Patients received the macrolide antibiotic roxithromycin 300 mg once daily for 28 days (506), or placebo (504). Primary endpoint was frequency of restenosis (diameter stenosis >50%) at follow-up angiography, and secondary endpoint was rate of target vessel revascularisation during the year after stenting. A prespecified secondary analysis addressed treatment effect with respect to titre of C pneumoniae in serum. Analysis was by intention to treat. FINDINGS: Rate of angiographic restenosis was 31% (157 lesions) in the roxithromycin group and 29% (148) in the placebo group (relative risk 1.08 [95% CI 0.92-1.26]; p50.43), corresponding to a rate of target vessel revascularisation of 19% (120) and 17% (105), respectively (1.13 [0.95-1.36]; p50.30). The combined 1-year rates of death and myocardial infarction were 7% (36) in the roxithromycin group and 6% (30) in the placebo group (p50.45). We showed a significant interaction between treatment and C pneumoniae antibody titre (p50.038 for restenosis, p50.006 for revascularisation), favouring roxithromycin at high titres (adjusted odds ratios at a titre of 1/512 were 0.44 [0.19-1.06] and 0.32 [0.13-0.81], respectively). INTERPRETATION: Non-selective use of roxithromycin is inadequate for prevention of restenosis after coronary stenting. There is, however, a differential effect dependent on C pneumoniae titres. In patients with high titres, roxithromycin reduced the rate of restenosis.