12 clinical studies on the use of roxithromycin 300mg daily in adult respiratory tract infection were subject to meta-analysis. Nine of the studies were comparative and 5 were from the published literature. Auditable individual patient data were reviewed for 11 of the 12 studies.4297 patients with respiratory tract infection were enrolled in these studies and 384 (8.9%) hadHaemophilus influenzaeidentified as a causative pathogen. Of these patients, 331 were treated with roxithromycin: 268 with 150mg twice daily and 63 with 300mg once daily. 53 patients were treated with comparator antibiotics. The primary analysis was undertaken on 300 cases whereH. influenzaewas the sole pathogen identified: 253 of these cases were treated with roxithromycin and 47 with comparator antibiotics.On a ‘per protocol’ (PPA) basis, the overall satisfactory clinical and bacteriological response for all roxithromycin regimens was 87% [95% confidence interval (CI), 82 to 92%] and 83%, respectively. Clinical response on an ‘intention-to-treat’ (ITT) basis was 80% (CI: 75 to 85%) excluding studies enrolling only hospitalised patients, and 78% (CI: 73 to 83%) for all cases. Clinical response (ITT) was marginally lower at 76% (CI: 72 to 82%) if mixed infections were included.Comparator antibiotics were amoxicillin/clavulanic acid, cefaclor, doxycycline, erythromycin and cotrimoxazole. A satisfactory clinical response (ITT) was obtained in 70% of cases treated with comparator antibiotics (CI: 57 to 83%). Roxithromycin demonstrated a response rate similar to that of amoxicillin/clavulanic acid and significantly better than erythromycin (p = 0.03). Although the numbers were small, roxithromycin was significantly more effective for the treatment of pneumonia in the directly comparative studies, with 93% success, compared with only 53% for the comparator antibiotics (p = 0.02). The response rate was similar to that reported for clarithromycin.Current minimum inhibitory concentration and disc zone susceptibility breakpoints of ≤ 1 mg/L and ≥ 22mm underestimate the clinical efficacy of roxithromycin againstH. influenzae.A disc zone diameter susceptibility breakpoint of ≥ 10mm is predictive of a successful outcome in 78% of cases (95% CI: 71 to 85%). This correlates well with recent recommendations to adopt susceptibility breakpoints of ≥ 10mm and ≤ 16 mg/L when using National Committee for Clinical Laboratory Standards (NCCLS) methodology.The results of this analysis support the empirical use of roxithromycin in respiratory tract infections whereH. influenzaeis a potential pathogen. Future clinical studies of respiratory tract infection with roxithromycin should evaluate these new breakpoints.