Background: Haemophilus influenzae is commonly isolated from the airways of COPD patients. Antibiotic treatment may cause the emergence of resistant H. influenzae strains, particularly ampicillin-resistant strains, including β-lactamase-negative ampicillin resistance (BLNAR) strains. Genetic identification using ftsI sequencing is the optimum method for identifying mutations within BLNAR strains. The prevalence of BLNAR in COPD patients during the stable state has not been reported. We investigated the antibiotic resistance patterns of H. influenzae present in the sputum of stable COPD patients, focusing on ampicillin resistance; the prevalence of enzyme and non-enzyme-mediated ampicillin resistance was determined. A subset of patients was followed up longitudinally to study H. influenzae strain switching and antibiotic sensitivity changes. Patients and methods: Sputum sampling was performed in 61 COPD patients, with 42 samples obtained at baseline; H. influenzae was detected by polymerase chain reaction in 28 samples. In all, 45 patients completed the follow-up for 2 years; 24 H. influenzae isolates were obtained. Results: Disk diffusion showed the highest antibiotic resistance in the penicillin antibiotic group (eg, 67% for ampicillin) and macrolides (eg, 46% for erythromycin), whereas all isolates were susceptible to quinolones. Of the 16 isolates resistant to ampicillin, 9 (56%) were β-lactamase positive. The β-lactamase-negative isolates were further investigated; none of these fulfilled the phenotypic BLNAR classification criteria of ampicillin minimum inhibitory concentration >1 µg/mL, and only one demonstrated an ftsI mutation. Frequent H. influenzae strain switching was confirmed using multilocus sequence typing and was associated with changes in the antibiotic sensitivity pattern. Conclusion: We observed an overidentification of ampicillin resistance by disk diffusion. The majority of ampicillin resistance was due to enzyme production. H. influenzae strain changes during the stable state may be associated with a change in antibiotic sensitivity; this has implications for empirical antibiotic prescribing.