Inappropriate benzodiazepine use continues to cause substantial morbidity and mortality globally. We aimed to characterise the initiation of new benzodiazepine treatment episodes in Australia and identify correlates of potentially inappropriate benzodiazepine use. We conducted a population-based cohort study using dispensing claims from a 10% sample of Pharmaceutical Benefit Scheme eligible Australians (2014-2017). Our cohort comprised adults initiating a new benzodiazepine treatment; we defined potentially inappropriate use as ≥3 benzodiazepine dispensing over any continuous 90-day period in the year following initiation. We examined characteristics associated with potentially inappropriate benzodiazepine use using multivariable logistic regression. People initiating a new benzodiazepine treatment episode (n = 276 765) were more frequently female (59.1%) and <65 years of age (73.6%). In the 90 days prior to initiating benzodiazepine, people were commonly dispensed antidepressants (26.5%), opioid analgesics (17.6%) and antipsychotics (4.7%). In the first year after initiation, 20 938 (9.5%) people experienced 'potentially inappropriate use'. Having a greater initial quantity of benzodiazepine dispensed [odds ratio (OR), 1.10; 95% confidence interval (CI) 1.08-1.12 per 10 defined daily doses increase], dispensing of antipsychotics (OR 3.00, 95% CI 2.86-3.15) and >5 unique medicines (OR 2.54, 95% CI 2.44-2.64; vs. ≤5 unique medicines) in the 90 days prior to initiation were associated with potentially inappropriate benzodiazepine use. Approximately, 1 in 10 people who initiated benzodiazepines were using it beyond the guideline recommended period. We identified factors at the time of initiation associated with potentially inappropriate use; clinicians should consider these before prescribing benzodiazepines and initiate conversations about alternative therapy when necessary.