OBJECTIVE: We examined changes in patterns of benzodiazepine use in British Columbia over a period of increasing evidence of harms associated with long-term use. METHODS: Using linked administrative databases for the years 1996 and 2006, we performed logistic regression to examine how socio-economic and health factors affect the likelihood of benzodiazepine use and long-term use, and to test for changes in rates of use and long-term use over time. RESULTS: In 2006, 8.4% of British Columbians used benzodiazepines, 3.5% long-term. Use was positively related with being female, lower income, older, and of poorer health status. Long-term use was positively associated with being in the lowest income quintile, of poorest health, and over the age of 65. While the rate of long-term use decreased from 1996 to 2006 for those over age 70, it increased in middle-aged populations. CONCLUSIONS: Our results suggest, despite increased awareness of and cautions regarding risks associated with long-term use of benzodiazepines, rates of potentially inappropriate use have changed very little over a decade. Given that early use of benzodiazepines is positively associated with later long-term use, policies targeting populations younger than conventionally studied (i.e. those under age 65) may be needed to decrease rates of long-term use.