BACKGROUND: Long-acting muscarinic antagonists (LAMA), long-acting beta2-agonists (LABA) and fixed dose combinations (FDC) of inhaled corticosteroids (ICS) and LABA are used as inhaled maintenance therapies for COPD. OBJECTIVE: To estimate persistece rates from dispensing patterns of long-acting inhaled drugs for COPD. METHODS: From the PHARMO-database, COPD patients starting LAMA, LABA or LABA-ICS FDC between 2002 and 2006 were selected. Persistence with the initial as well as with any long-acting inhaled drug was determined, defined as time between start and stop of initial/any therapy, allowing </= 60-days gaps between refills. For patients who did not continue to receive dispensings of the initial therapy for at least one year, the first change in therapy was determined. RESULTS: The study included 2201 LAMA, 1201 LABA and 4146 LABA-ICS FDC users. Persistence rates with initial therapy alone at 1, 2, and 3 years were 25%, 14%, 8% for LAMA, 21%, 10%, 6% for LABA and 27%, 14%, 8% for LABA-ICSFDC. Of patients who did not persist with LAMA alone for one year, 15% added and 13% switched therapy (both mostly LABA-ICS FDC). Of patients not persisting with LABA alone, 9% added therapy (mostly LAMA) and 31% switched therapy (mostly to LABA-ICS FDC). In patients not persisting with LABA-ICS FDC, add-on and switch occurred equally frequent (11%, mostly LAMA). Persistence rates with any long-acting drug at 1, 2 and 3 years were 36%, 23% and 17% respectively. CONCLUSION: Persistence with the initial as well as with any long-acting inhaled drug in COPD is low, with a substantial proportion of patients changing therapy.