The long-term outcome time to relapse determined as the exacerbation-free interval (EFI) has been proposed as a standard measure for comparing the efficacy of antimicrobial therapies in acute exacerbation of chronic obstructive bronchitis (AECOB). In this 6-month, randomised, open-label study, the efficacy of 10 days of oral levofloxacin 500 mg once daily or cefuroxime 250 mg twice daily was evaluated in 689 well-defined patients experiencing AECOB episodes. In the clinically evaluable per-protocol (PPc) population and the modified intent-to-treat population, the clinical cure rates at test of cure were, respectively, 94.6% for levofloxacin versus 93.8% for cefuroxime (0.8% difference, 95% confidence interval (CI) -3.2 to 4.8) and 94.5% for levofloxacin versus 92.2% for cefuroxime (2.3% difference, 95% CI -1.8 to 6.2), whilst the probability that 25% of patients would relapse during follow-up was reached within 93 days for levofloxacin compared with 81 days for cefuroxime in the PPc population (P=0.756). A multivariate analysis revealed that only congestive heart failure and number of AECOB episodes in the previous 12 months were predictive of relapse. Safety was comparable in the two treatment groups, with possibly related treatment-emergent adverse events occurring in 5.0% and 2.9% of subjects in the levofloxacin and cefuroxime groups, respectively. In addition to demonstrating the non-inferiority of levofloxacin compared with cefuroxime in AECOB, the data from this study raise the question of whether EFI is a useful discriminative endpoint for comparing antimicrobial therapies.