OBJECTIVES: Current clinical guidelines discourage long-term prescription of benzodiazepines and Z-drugs (BZD); however, the practice continues to exist. The aim of this study was to investigate the proportion of long-term BZD prescriptions and its risk factors. DESIGN: Retrospective cohort study using a health insurance database.Japan. PARTICIPANTS: A total of 86 909 patients were identified as outpatients aged 18 to 65 years who started BZD between 1 October 2012 and 1 April 2015. After excluding patients who underwent surgery on the day of first BZD prescription (n=762) and patients without 8 months follow-up (n=12 103), 74 044 outpatients were analysed. MAIN OUTCOME MEASURES: We investigated the proportion of long-term prescriptions for ≥8 months among new BZD users. We assessed patient demographics, diagnoses, characteristics of the initial BZD prescription and prescribers as potential predictors of the long-term BZD prescription. Multivariable logistic regression was performed to assess the association between long-term prescription and potential predictors. RESULTS: Of the new BZD users, 6687 (9.0%) were consecutively prescribed BZD for ≥8 months. The long-term prescription was significantly associated with mood and neurotic disorder, cancer, prescription by psychiatrists, multiple prescriptions, hypnotics and medium half-life BZD in the initial prescription. CONCLUSION: Despite the recent clinical guidelines, 9% of new BZD users were given prescriptions for more than 8 months. Physicians should be aware of risk factors when prescribing BZDs for the first time.